A real-time monitoring, analysis and early warning system for sludge treatment equipment

Abstract

The present invention relates to the technical field of equipment monitoring and early warning, and discloses a real-time monitoring, analysis and early warning system for sludge treatment equipment, including: a behavior slice construction module, configured to collect equipment status data based on preset key behavior event types to form behavior slices; a behavior slice processing module, configured to construct a coupled feature vector including response stability features, dynamic tension impact features and coupled disturbance rejection ability features; a behavior map generation module, configured to aggregate the current coupled feature vector and the coupled feature vectors of the already constructed behavior slices to form a behavior map; a risk score generation module, configured to match the current behavior map with a state cluster template to calculate the risk score of the current equipment; and a risk early warning module, configured to determine the risk level according to the risk score and output corresponding early warning measures. The present invention realizes real-time monitoring of the operation status of sludge treatment equipment, pre-identification of fault risks and multi-level early warning linkage.

Description

Technical Field

[0001] The present invention belongs to the technical field of equipment monitoring and early warning, and particularly relates to a real-time monitoring, analysis and early warning system for sludge treatment equipment. Background Art

[0002] As an important link in the sewage treatment process, sludge treatment widely depends on the stable operation of core mechanical equipment such as centrifugal dehydrators and belt filter presses. The equipment operates under high load and continuous working conditions, often accompanied by phenomena such as mechanical shock, load fluctuation, wear and aging. If abnormal operation is not identified in time, it is very easy to cause problems such as equipment shutdown and reduced treatment capacity, thereby affecting the efficiency and safety of the entire sewage treatment system.

[0003] Current common equipment monitoring methods mainly include regular manual inspections and simple alarm systems based on single-parameter thresholds. For example, fixed upper and lower limits are set through basic sensors such as current, voltage, and temperature to achieve over-limit alarm prompts. However, such methods have significant limitations: on the one hand, the manual inspection cycle is long and the coverage is limited, which easily leads to the late discovery of key states; on the other hand, single-point monitoring is difficult to reflect the multi-dimensional operation characteristics and behavior evolution process of the equipment, and it is impossible to effectively identify potential risks such as bearing fatigue, filter belt relaxation, and abnormal sludge inlet.

[0004] In addition, traditional early warning systems lack the ability to analyze the associated changes of multiple parameters and are difficult to form an overall understanding of the equipment state. They often only trigger an alarm after obvious equipment failures or performance mutations occur, resulting in a lag in the early warning effect and easily missing the best intervention opportunity. Summary of the Invention

[0005] The present invention provides a real-time monitoring, analysis and early warning system for sludge treatment equipment, which solves the technical problems of late early warning, difficulty in realizing multi-parameter fusion and early fault identification in related technologies.

[0006] The present invention provides a real-time monitoring, analysis and early warning system for sludge treatment equipment, including:

[0007] A behavior slice construction module, which is used to, during the operation of the equipment, based on preset key behavior event types, real-time monitor the equipment state data. When a key behavior event that meets the event trigger condition is detected, record the trigger time, and collect the equipment state data within a first preset time period before and after the trigger time to form a behavior slice, and represent the equipment state data within the first preset time period through a first feature sequence;

[0008] A behavior slice processing module, which is used to process the first feature sequence in the behavior slice to construct a coupled feature vector including response stability features, dynamic tension impact features, and coupled anti-interference ability features;

[0009] The behavior graph generation module is used to aggregate the coupling feature vectors corresponding to the current behavior slice and the coupling feature vectors of the behavior slices already constructed during the current device operation to form a behavior graph, where the behavior graph is a two-dimensional matrix structure composed of coupling feature vectors, and each row corresponds to the coupling feature vector of a behavior slice;

[0010] The risk score generation module is used to match the current behavior graph with the state cluster templates in the pre-established historical graph database, calculate the structural distance between the current behavior graph and each state cluster template, select the state cluster template with the highest similarity, determine the state cluster category of the current behavior graph, and calculate the risk score of the current device according to the risk score interval corresponding to the state cluster category and the matching deviation degree between the current behavior graph and the state cluster template;

[0011] The risk warning module is used to determine the risk level of the device according to the risk score and output the corresponding warning measures.

[0012] Further, the device state data includes: current, vibration acceleration, temperature, pressure, rotational speed, and voltage.

[0013] Further, the difference between the maximum value and the minimum value of the vibration acceleration in the first feature sequence of the behavior slice is multiplied by the sum of the standard deviation of the current and the standard deviation of the temperature to obtain the first intermediate feature; the first intermediate feature is combined with the product of the average value of the current and the average value of the temperature to obtain the response stability feature;

[0014] The square of the difference between the maximum value and the minimum value of the vibration acceleration in the first feature sequence of the behavior slice is used to obtain the second intermediate feature, the square of the average value of the current and the square of the average value of the temperature are used to obtain the third intermediate feature, the ratio of the standard deviation of the rotational speed to the average value of the rotational speed is used to obtain the fourth intermediate feature, and the second intermediate feature, the third intermediate feature, and the fourth intermediate feature are combined to obtain the dynamic tension impact feature;

[0015] The square of the difference between the maximum value and the minimum value of the current in the first feature sequence of the behavior slice is used to obtain the fifth intermediate feature, the square of the difference between the maximum value and the minimum value of the temperature is used to obtain the sixth intermediate feature, the sum of the square of the average value of the vibration acceleration and the square of the standard deviation of the rotational speed is used to obtain the seventh intermediate feature, and the fifth intermediate feature, the sixth intermediate feature, and the seventh intermediate feature are combined to obtain the coupling anti-interference ability feature.

[0016] Further, the construction method of the behavior graph includes:

[0017] Select the behavior slices in the preset time window according to the first preset rule in the time order of the occurrence of the behavior slices;

[0018] The coupling eigenvectors corresponding to each row of slices are weighted according to the weight mechanism and arranged in chronological order to form a two-dimensional matrix structure.

[0019] Further, the first preset rule is used to automatically adjust the selection method of the time window according to the current operation stage of the device;

[0020] Among them, the operation stage includes: startup stage, stable operation stage, and shutdown transition stage;

[0021] For the startup stage, all the behavior slices within the second preset time period are selected;

[0022] For the stable operation stage, the behavior slices whose time interval from the current slice is within the third preset time period are selected;

[0023] For the shutdown transition stage, all the behavior slices of the current operation cycle are selected.

[0024] Further, the operation stage is determined according to the deviation rate between the real-time speed and the set speed of the device. Among them, the deviation rate is obtained by the ratio of the absolute value of the difference between the real-time speed and the set speed to the set speed. The determination method of the operation stage includes:

[0025] When the deviation rate is greater than or equal to the first preset threshold and the change rate of the deviation rate is positive, it is determined to be the startup stage;

[0026] When the deviation rate is less than the first preset threshold, it is determined to be the stable operation stage;

[0027] When the deviation rate is greater than or equal to the first preset threshold and the change rate of the deviation rate is negative, it is determined to be the shutdown transition stage.

[0028] Further, the weight mechanism is: according to the time interval between the behavior slice time and the current time and the mean time between failures, introduce a behavior time gain coefficient, and calculate the weight in an exponential decay manner. The behavior event gain coefficient is set according to the key behavior event type.

[0029] Further, in the matching process between the behavior graph and the state cluster template in the risk scoring generation module, a distance calculation method that combines multiple similarity measurement indicators is adopted. The distance calculation method includes the following three indicators:

[0030] Matrix structure consistency index, which is calculated by constructing a sliding window on the behavior graph and the state cluster template, respectively calculating the mean and standard deviation of the features within the sliding window, and calculating based on the difference of the corresponding statistical features between the sliding windows;

[0031] The dynamic time series alignment distance metric forms a sequence by extracting the coupled feature vectors in the behavior graph and the state cluster template, uses the dynamic time warping algorithm to perform non-linear alignment on it, and calculates the length of the minimum alignment path to obtain the dynamic time series alignment distance metric;

[0032] The feature space distance metric calculates the distribution deviation of the behavior graph and the state cluster template in the multi-dimensional feature space using the Mahalanobis distance;

[0033] By performing weighted fusion on the three metrics, the structural distance is obtained.

[0034] Furthermore, the risk scoring generation module performs linear mapping within the risk scoring interval corresponding to the state cluster template based on the structural distance between the current behavior graph and the state cluster template to obtain the risk score.

[0035] The beneficial effects of the present invention are as follows: The present invention adopts a slicing mechanism driven by key behavior events, which can actively capture the non-linear dynamic response during the operation of the device. Through multi-parameter synchronous acquisition and behavior feature extraction, it realizes the real-time identification of the device state and the modeling of the behavior evolution trend, makes up for the deficiency of the traditional timed sampling warning lag, and significantly improves the sensitivity and accuracy of early fault identification;

[0036] The present invention extracts multi-dimensional features such as response stability, dynamic tension impact, and coupling anti-interference ability, constructs a behavior graph, and combines the operation stage adaptive aggregation strategy and the differential weight mechanism to comprehensively reflect the behavior evolution mode of the device at different time periods, enhances the system's understanding ability of complex working condition changes, and provides high-value input for subsequent risk identification;

[0037] The present invention determines the state cluster category of the current behavior graph through the structural distance matching method, performs linear mapping within the risk scoring interval, and outputs a continuous risk score; then combines the scoring level interval to output corresponding warning measures, realizes the linkage of risk trend identification, level division, and response strategy, and greatly improves the active warning ability and operation and maintenance efficiency of the sludge treatment equipment. Brief Description of the Drawings

[0038] Figure 1 It is a module schematic diagram of a real-time monitoring, analysis and warning system for a sludge treatment equipment of the present invention. Detailed Embodiments

[0039] Reference will now be made to example embodiments to discuss the subject matter described herein. It should be understood that discussing these embodiments is only to enable those skilled in the art to better understand and thus implement the subject matter described herein. Without departing from the scope of protection of the content of this specification, changes can be made to the functions and arrangements of the elements discussed. Each example can omit, substitute, or add various processes or components as needed. Additionally, the features described relative to some examples can also be combined in other examples.

[0040] As Figure 1 shown, a real-time monitoring, analysis and early warning system for sludge treatment equipment includes:

[0041] A behavior slice construction module 101, configured to, during the operation of the equipment, based on a preset key behavior event type, monitor the equipment status data in real time. When a key behavior event that meets the event trigger condition is detected, record the trigger time, and collect the equipment status data within a first preset time period before and after the trigger time to form a behavior slice, and represent the equipment status data within the first preset time period through a first feature sequence;

[0042] A behavior slice processing module 102, configured to process the first feature sequence in the behavior slice to construct a coupled feature vector including response stability features, dynamic tension impact features, and coupled anti-interference ability features;

[0043] A behavior map generation module 103, configured to aggregate the coupled feature vector corresponding to the current behavior slice and the coupled feature vectors of the behavior slices constructed during the current equipment operation to form a behavior map, where the behavior map is a two-dimensional matrix structure composed of coupled feature vectors, and each row corresponds to the coupled feature vector of a behavior slice;

[0044] A risk score generation module 104, configured to match the current behavior map with the state cluster templates in a pre-established historical map database, calculate the structural distance between the current behavior map and each state cluster template, select the state cluster template with the highest similarity, determine the state cluster category of the current behavior map, and calculate the risk score of the current equipment according to the risk score interval corresponding to the state cluster category and the matching deviation degree between the current behavior map and the state cluster template;

[0045] A risk early warning module 105, configured to determine the risk level of the equipment according to the risk score and output corresponding early warning measures.

[0046] In an embodiment of the present invention, the equipment status data includes: current, vibration acceleration, temperature, pressure, rotational speed, and voltage.

[0047] Among them, the current and voltage are collected through a current sensor and a voltage sensor respectively. The unit of current is A, the unit of voltage is V, the vibration acceleration is collected by a three-axis accelerometer, and the unit is , the temperature is collected by an infrared sensor, the unit is °C, the pressure is collected by a pressure sensor, the unit is MPa, and the rotational speed is collected by a Hall encoder, the unit is RPM;

[0048] In this embodiment, the maximum-minimum normalization method is used to normalize the device status data to ensure the comparability and calculation stability among different physical quantities in the coupling feature calculation.

[0049] In an embodiment of the present invention, the types of key behavior events include: startup, shutdown, load mutation, temperature anomaly, sludge inlet fluctuation, and manual control instructions, etc. Among them, the trigger condition for load mutation is: the load change rate per unit time exceeds the first preset threshold; the trigger condition for temperature anomaly is: the rate of temperature increase per unit time exceeds the second preset threshold; the trigger condition for sludge inlet fluctuation is: the change amount of the sludge inlet flow per unit time exceeds the third preset threshold.

[0050] The data acquisition method in this embodiment is different from the traditional timing sampling mechanism, and has an event-driven data acquisition logic, which can improve the recognition ability of the device's non-linear response and enhance the accuracy and response efficiency of real-time warning.

[0051] In an embodiment of the present invention, the calculation formula for the response stability feature is:

[0052] , where represents the response stability feature, which is used to measure whether the vibration amplitude and current fluctuation of the device are amplified synchronously with the temperature fluctuation, and whether these fluctuations are abnormally high relative to the power, and respectively represent the maximum and minimum values of the vibration acceleration in the first feature sequence, represents the standard deviation of the current in the first feature sequence, represents the standard deviation of the temperature in the first feature sequence, represents the average value of the current in the first feature sequence, represents the average value of the temperature in the first feature sequence;

[0053] In an embodiment of the present invention, the calculation formula for the dynamic tension impact feature is:

[0054] , where Indicates the dynamic tension impact feature, which is used to measure whether the device has an excessive response to disturbances. Suitable scenarios include: unstable filter belt tension; mechanical system instability under frequent speed regulation of the frequency converter; local jamming, looseness, and fatigue impact of bearings or rollers; log represents the logarithmic operation, Represents the standard deviation of the rotational speed in the first feature sequence, Represents the average value of the rotational speed in the first feature sequence;

[0055] In an embodiment of the present invention, the calculation formula for the coupling anti-disturbance ability feature is:

[0056] , where, Represents the coupling anti-disturbance ability feature, which is used to measure whether the device has good anti-disturbance performance under current fluctuations and temperature fluctuations, and respectively represent the maximum and minimum values of the current in the first feature sequence, and respectively represent the maximum and minimum values of the temperature in the first feature sequence, Represents the average value of the vibration acceleration in the first feature sequence, Represents the standard deviation of the rotational speed in the first feature sequence.

[0057] Arrange the response stability feature, dynamic tension impact feature, and coupling anti-disturbance ability feature in a set order to construct the coupling feature vector corresponding to the behavior slice.

[0058] In an embodiment of the present invention, the construction method of the behavior map includes:

[0059] Select the behavior slices in the preset time window according to the first preset rule in the chronological order of the occurrence of the behavior slices;

[0060] Weight the coupling feature vectors corresponding to each behavior slice according to the weight mechanism and arrange them in chronological order to form a two-dimensional matrix structure.

[0061] In an embodiment of the present invention, the first preset rule is used to automatically adjust the selection method of the time window according to the current operating stage of the device;

[0062] where the operating stage includes: start-up stage, stable operation stage, and shutdown transition stage;

[0063] For the start-up stage, select all behavior slices within the second preset time period;

[0064] For the stable operation stage, select the behavior slices whose time interval from the current slice is within the third preset time period;

[0065] For the shutdown transition stage, all behavior slices of the current operation cycle are selected.

[0066] Preferably, the second preset time period is set to 10 minutes, and the third preset time period is set to 3 minutes; the first preset rule can automatically match the graph window range according to the operation stage of the device, improving the adaptability and representativeness of the behavior graph to the current state.

[0067] In an embodiment of the present invention, the operation stage is determined according to the deviation rate between the real-time speed and the set speed of the device, where the deviation rate is obtained by the ratio of the difference between the real-time speed and the set speed to the set speed, and the calculation formula of the deviation rate is: , represents the deviation rate, represents the real-time speed of the device, represents the set speed;

[0068] The determination method of the operation stage includes:

[0069] When the deviation rate is greater than or equal to the first preset threshold and the change rate of the deviation rate is positive, it is determined to be the startup stage;

[0070] When the deviation rate is less than the first preset threshold, it is determined to be the stable operation stage;

[0071] When the deviation rate is greater than or equal to the first preset threshold and the change rate of the deviation rate is negative, it is determined to be the shutdown transition stage. Preferably, the first preset threshold is 10%.

[0072] In an embodiment of the present invention, the weight mechanism is as follows: According to the time interval between the behavior slice time and the current time and the mean time between failures, a behavior time gain coefficient is introduced, and the weight is given in an exponential decay manner. The behavior event gain coefficient is set according to the key behavior event type, and the calculation formula of the weight is: , W represents the weight, represents the time interval from the current time, MTBF represents the mean time between failures, represents the behavior event gain coefficient, which is preset according to the key behavior event type. For example, for the load mutation type, the behavior event gain coefficient takes a value of 2.5, and for the startup type, the behavior event takes a value of 1.5.

[0073] In an embodiment of the present invention, different behavior event gain coefficients are introduced for different key behavior event types , The numerical values are used to reflect the importance of such events to the operating state of the device. Events with a high correlation with potential risks or sudden failures, such as load mutation and temperature anomaly, are set with a higher coefficient; events representing operation behaviors, such as startup and manual control instructions, are set with a medium coefficient; routine process events, such as sludge inlet fluctuation and shutdown, are set with a basic coefficient. In this way, the weight expression of key behavior responses in the behavior graph is enhanced, and the sensitivity and accuracy of overall fault prediction are improved.

[0074] In an embodiment of the present invention, in the matching process between the behavior graph and the state cluster template in the risk score generation module, a distance calculation method that combines multiple similarity measurement metrics is adopted. The distance calculation method includes the following three metrics:

[0075] Matrix structure consistency metric, which is obtained by constructing a sliding window on the behavior graph and the state cluster template, respectively calculating the mean and standard deviation of the features within the sliding window, and calculating based on the difference of the corresponding statistical features between the sliding windows;

[0076] Dynamic time series alignment distance metric, which is obtained by extracting the sequences formed by the coupled feature vectors in the behavior graph and the state cluster template, using the dynamic time warping algorithm to perform non-linear alignment on them, and calculating the length of the minimum alignment path;

[0077] Feature space distance metric, which uses the Mahalanobis distance to calculate the distribution deviation of the behavior graph and the state cluster template in the multi-dimensional feature space;

[0078] By weighted fusion of the three metrics, the structural distance is obtained. The calculation formula of the structural distance is:

[0079] , where D represents the structural distance, represents the matrix structure consistency metric, represents the dynamic time series alignment distance metric, represents the feature space distance metric, , and respectively represent the first weight coefficient, the second weight coefficient, and the third weight coefficient.

[0080] Specifically, the matrix structure consistency metric is used to measure the similarity degree of the two-dimensional arrangement structures of the behavior graph and the state cluster template; in the present invention, sliding windows are constructed on the behavior graph and the state cluster template in the same way, and the mean and standard deviation of each dimension of the coupled features within each window area are respectively calculated, and then the statistical feature values under the corresponding windows are compared to obtain the difference between the windows. The matrix structure consistency metric is composed of the comprehensive difference of all windows. This method can effectively capture the local response patterns of the behavior stages and their spatial distribution differences.

[0081] The dynamic time series alignment distance metric is used to evaluate the consistency of the evolution trends of the behavior graph and the state cluster template in the time dimension. First, the invention extracts the coupled feature vectors of each row in the behavior graph and the state cluster template in chronological order to form two time series, and uses the dynamic time warping algorithm to perform non-linear alignment on these two time series, allowing for local time stretching and shrinking. Finally, the shortest path length of the alignment path is used as the dynamic time series alignment distance metric; this metric can capture the structural similarity of the same type of operating states under different response times.

[0082] The feature space distance metric is used to judge the degree of distribution deviation of the behavior graph and the state cluster template as a whole in the high-dimensional coupled feature space. The invention calculates it using the Mahalanobis distance. By calculating the covariance between the features of each dimension in the behavior graph and the state cluster template, the similarity of the behavior graph and the state cluster template in the statistical distribution can be accurately evaluated. The Mahalanobis distance has good adaptability to the correlation between features and can avoid the problem of repeated amplification of highly correlated features by the traditional Euclidean distance.

[0083] The invention weights and fuses the above three similarity metrics according to the preset weight coefficients to form a unified structural distance. This structural distance is used to match the state cluster template closest to the current behavior graph, and combines its risk score interval and deviation degree to generate a final score, which is used as the basis for risk determination of the device operating state.

[0084] In an embodiment of the invention, the risk score generation module performs a linear mapping within the risk score interval corresponding to the state cluster template based on the structural distance between the current behavior graph and the state cluster template to obtain the risk score. The calculation formula for the risk score is: , represents the risk score, and represent the lower and upper limits of the risk score interval corresponding to the state cluster template, represents the maximum acceptable structural distance in the state cluster template, obtained through data statistics.

[0085] The state cluster template is generated through cluster analysis based on a large number of historical equipment behavior graphs. Each state cluster template corresponds to a state cluster category. Each state cluster template is associated with a preset risk score interval, which is used to indicate the risk floating range under this type of operating state. When the behavior graph matches the state cluster template, the present invention uses the structural distance between the behavior graph and the state cluster template as the basis for score adjustment, and performs proportional mapping within the score interval: when the structural distance is small, the score value is close to the lower limit of the interval, indicating a lower risk; when the structural distance is large, the score value is close to the upper limit of the interval, indicating a higher risk. This scoring method realizes the continuous quantification of the risk level, can more finely reflect the degree of deviation between the equipment operating state and the standard template, and enhances the accuracy and response sensitivity of the scoring results.

[0086] In one embodiment of the present invention, the risk warning module determines the risk level of the current operating state of the equipment according to the risk score, and outputs corresponding warning measures according to the risk level. The risk level is divided according to the numerical range of the risk score. Preferably, the score value is divided into three level intervals:

[0087] Low risk level: When the risk score is in the first score range (for example, 0-30), the device status is determined to be normal, and only the operation log is recorded without manual intervention;

[0088] Medium risk level: When the risk score is in the second scoring range (e.g. 31-60), an early warning message will be issued to remind the operation and maintenance personnel to pay attention to the equipment operation status and recommend a preliminary inspection;

[0089] High risk level: When the risk score is in the third scoring range (for example, 61-100), a serious alarm will be triggered, including warning light signals, interface pop-up windows, SMS or APP notifications, and can be linked to trigger subsequent control strategies, such as speed limit, load reduction, shutdown protection and other operations.

[0090] This risk level classification mechanism can output clear level judgment results based on the continuous numerical value of the risk score, and link with actual early warning measures to ensure that equipment can receive attention in the budding stage of risks and can respond quickly in high-risk stages, thereby reducing the probability of sudden failures and operating losses.

[0091] In one embodiment of the present invention, the present invention supports remote real-time monitoring of the equipment status data of all sludge treatment equipment, and after preliminary local processing, it is wirelessly transmitted to the cloud platform. Users can remotely access it through the software system on the web or mobile terminal to achieve unified monitoring and centralized management of multiple projects and multiple equipment. This deployment mode improves the scalability and remote operation and maintenance capabilities of the system, and provides data foundation support for graph analysis and risk scoring mechanisms.

[0092] It should be noted that the setting of the interval and the threshold value is for the convenience of comparison. The size of the threshold value depends on the amount of sample data and the base quantity set by those skilled in the art for each group of sample data, as long as it does not affect the proportional relationship between the parameters and the quantized values. And the above formulas are all calculations of taking the numerical values without dimensions. The formulas are all obtained by collecting a large amount of data for software simulation to get a formula closest to the actual situation. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0093] The above has described the embodiments of the present invention. However, the present invention is not limited to the above specific embodiments. The above specific embodiments are merely illustrative rather than restrictive. Under the inspiration of this embodiment, those of ordinary skill in the art can also make many forms, all of which fall within the protection scope of this embodiment.

Claims

1. A real-time monitoring, analysis and early warning system for sludge treatment equipment, characterized in that, Including: A behavior slice construction module, which is used to, during the operation of the device, monitor the device status data in real time based on preset key behavior event types. When a key behavior event that meets the event trigger condition is detected, record the trigger time, and collect the device status data within the first preset time period before and after the trigger time to form a behavior slice, and represent the device status data within the first preset time period through a first feature sequence; A behavior slice processing module, which is used to process the first feature sequence in the behavior slice to construct a coupled feature vector including response stability features, dynamic tension impact features, and coupled anti-interference ability features; Among them, the difference between the maximum value and the minimum value of the vibration acceleration in the first feature sequence of the behavior slice is multiplied by the sum of the standard deviation of the current and the standard deviation of the temperature to obtain a first intermediate feature; the first intermediate feature is combined with the product of the average value of the current and the average value of the temperature to obtain a response stability feature; The square of the difference between the maximum value and the minimum value of the vibration acceleration in the first feature sequence of the behavior slice is used to obtain a second intermediate feature, the square of the average value of the current and the square of the average value of the temperature are used to obtain a third intermediate feature, and the ratio of the standard deviation of the rotational speed to the average value of the rotational speed is used to obtain a fourth intermediate feature. The second intermediate feature, the third intermediate feature, and the fourth intermediate feature are combined to obtain a dynamic tension impact feature; The square of the difference between the maximum value and the minimum value of the current in the first feature sequence of the behavior slice is used to obtain a fifth intermediate feature, the square of the difference between the maximum value and the minimum value of the temperature is used to obtain a sixth intermediate feature, and the sum of the square of the average value of the vibration acceleration and the square of the standard deviation of the rotational speed is used to obtain a seventh intermediate feature. The fifth intermediate feature, the sixth intermediate feature, and the seventh intermediate feature are combined to obtain a coupled anti-interference ability feature; A behavior map generation module, which is used to aggregate the coupled feature vector corresponding to the current behavior slice and the coupled feature vectors of the behavior slices constructed during the current device operation to form a behavior map. Among them, the behavior map is a two-dimensional matrix structure composed of coupled feature vectors, and each row corresponds to the coupled feature vector of a behavior slice; A risk score generation module, which is used to match the current behavior map with the state cluster templates in the pre-established historical map database, calculate the structural distance between the current behavior map and each state cluster template, select the state cluster template with the highest similarity, determine the state cluster category of the current behavior map, and calculate the risk score of the current device according to the risk score interval corresponding to the state cluster category and the matching deviation degree between the current behavior map and the state cluster template; A risk warning module, which is used to determine the risk level of the device according to the risk score and output corresponding warning measures.

2. The real-time monitoring, analysis and early warning system for a sludge treatment device according to claim 1, characterized in that The device status data includes: current, vibration acceleration, temperature, pressure, rotational speed, and voltage.

3. The real-time monitoring, analysis and early warning system for a sludge treatment device according to claim 1, characterized in that, The construction method of the behavior map includes: Select the behavior slices of the preset time window according to the first preset rule in the time order of the occurrence of the behavior slices; The coupled feature vectors corresponding to each behavior slice are weighted according to the weight mechanism and arranged in time order to form a two-dimensional matrix structure.

4. The real-time monitoring, analysis and warning system for a sludge treatment device according to claim 3, characterized in that, The first preset rule is used to automatically adjust the selection method of the time window according to the current operation stage of the device; Among them, the operation stage includes: startup stage, stable operation stage, and shutdown transition stage; For the startup stage, all behavior slices within the second preset time period are selected; For the stable operation stage, the behavior slices whose time intervals from the current slice are within the third preset time period are selected; For the shutdown transition stage, all behavior slices of the current operation cycle are selected.

5. The real-time monitoring, analysis and early warning system for a sludge treatment device according to claim 4, wherein, The operation stage is determined according to the deviation rate between the real-time speed and the set speed of the device. Among them, the deviation rate is obtained by the ratio of the absolute value of the difference between the real-time speed and the set speed to the set speed. The determination method of the operation stage includes: When the deviation rate is greater than or equal to the first preset threshold and the change rate of the deviation rate is positive, it is determined as the startup stage; When the deviation rate is less than the first preset threshold, it is determined as the stable operation stage; When the deviation rate is greater than or equal to the first preset threshold and the change rate of the deviation rate is negative, it is determined as the shutdown transition stage.

6. The real-time monitoring, analysis and early warning system for a sludge treatment device according to claim 3, wherein, The weight mechanism is as follows: According to the time interval between the behavior slice time and the current time and the mean time between failures, a behavior time gain coefficient is introduced, and the weight is calculated in an exponential decay manner. The behavior event gain coefficient is set according to the key behavior event type.

7. A real-time monitoring, analysis and early warning system for a sludge treatment device according to claim 1, characterized in that, In the matching process of the behavior graph and the state cluster template in the risk score generation module, a distance calculation method that fuses multiple similarity measurement indicators is adopted. The distance calculation method includes the following three indicators: Matrix structure consistency index, which is obtained by constructing a sliding window on the behavior graph and the state cluster template, calculating the mean and standard deviation of the features within the sliding window respectively, and calculating based on the difference of the corresponding statistical features between the sliding windows; Dynamic time series alignment distance index, which is obtained by extracting the sequences formed by the coupled feature vectors in the behavior graph and the state cluster template, using the dynamic time warping algorithm to perform non-linear alignment on them, and calculating the length of the minimum alignment path; Feature space distance index, which uses the Mahalanobis distance to calculate the distribution deviation of the behavior graph and the state cluster template in the multi-dimensional feature space; By performing weighted fusion on the three indicators, the structure distance is obtained.

8. The real-time monitoring, analysis and early warning system for a sludge treatment device according to claim 1, characterized in that, The risk score generation module performs linear mapping within the risk score interval corresponding to the state cluster template based on the structure distance between the current behavior graph and the state cluster template to obtain the risk score.

Images