Intelligent fault prediction system and method for complex electromechanical system

Abstract

The invention discloses a complex electromechanical system fault intelligent prediction system and method, a data acquisition unit, an association processing unit, a fault analysis unit and a data display unit, and relates to the technical field of fault prediction. According to the intelligent fault prediction system and method for the complex electromechanical system, the reliability of the system is improved, the operation and maintenance cost is reduced, powerful decision support is provided for management personnel, and behavior modes of the system in normal and abnormal states can be identified by analyzing multiple operation parameters with different attributes and carrying out association processing on the parameters. Therefore, factors possibly causing system faults can be found in advance, so that preventive measures are taken, potential problems can be predicted before the faults occur, and the downtime of the system is avoided or reduced. Therefore, not only can production loss caused by sudden failure be reduced, but also the emergency maintenance cost can be reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of fault prediction, and specifically to an intelligent fault prediction system and method for complex electromechanical systems. Background Art

[0002] With the increasing complexity of industrial systems, traditional fault detection and maintenance methods are difficult to meet the needs of modern production. Complex electromechanical systems usually involve multiple interrelated parameters and components, making it very difficult to predict potential faults in a timely and accurate manner.

[0003] Existing fault prediction methods often rely on manual inspections and regular detections. This method is not only time-consuming and laborious, but may also miss some key data, resulting in the inability to detect abnormalities or predict faults in a timely manner; many existing technical solutions cannot provide real-time monitoring and dynamic analysis, which limits the ability to quickly respond to changes in system status and increases the risk of faults occurring; without efficient data processing and analysis tools, it is difficult to accurately identify fault signs from a large amount of complex operation data, resulting in low accuracy and efficiency of fault diagnosis.

[0004] Due to the lack of effective prediction and timely fault information, managers often take countermeasures only after a fault occurs. This passive approach is not conducive to reducing the losses caused by faults, nor to improving the overall reliability and stability of the system.

[0005] Therefore, the present invention aims to provide an intelligent fault prediction system for complex electromechanical systems, which can achieve efficient and accurate prediction of potential faults in electromechanical systems through advanced data acquisition, correlation processing, fault analysis, and data display technologies, thereby improving the reliability and operation and maintenance efficiency of the system. Summary of the Invention

[0006] Aiming at the deficiencies of the prior art, the present invention provides an intelligent fault prediction system and method for complex electromechanical systems, which solves the problems raised in the background art.

[0007] To achieve the above objectives, the present invention is realized through the following technical solutions: An intelligent fault prediction system for complex electromechanical systems, comprising: A data acquisition unit, configured to obtain multiple operation parameters with different attributes in the electromechanical system, A correlation processing unit, configured to perform correlation processing on the operation parameters with different attributes to obtain multiple abnormal state attribute standard sets; wherein, each abnormal state attribute standard set contains operation parameters corresponding to multiple different attributes, and they are in an associated relationship in this abnormal state; A fault analysis unit is used to obtain real-time operation parameters of multiple different attributes during the current operation cycle of the electromechanical system, extract and process abnormal data, obtain corresponding deviation multiples, and based on the respective attributes corresponding to the deviation multiples, analyze the signal results with each abnormal state attribute standard set, and obtain corresponding prediction results or abnormal unknown signals. A data display unit is used to display the corresponding prediction results and abnormal unknown signals during the current operation cycle of the electromechanical system to the management personnel.

[0008] Preferably, the operation parameters are obtained by sensor detection and include but are not limited to current parameters, temperature parameters, humidity parameters, and voltage parameters.

[0009] Preferably, the correlation processing method is as follows: Step1. When the electromechanical system is in a normal state in the early stage, obtain the operation parameters of multiple same attributes collected from the electromechanical system at multiple time nodes, then perform standard integration processing on the operation parameters of the same attribute, and obtain the standard parameters of each attribute representing the normal operation state of the electromechanical system. Step2. When multiple abnormal states occur in the electromechanical system in the early stage, obtain the operation parameters of multiple different attributes collected from the electromechanical system at the same time node. Step3. Compare the operation parameters of each attribute in each abnormal state with their corresponding standard parameter intervals: Select the operation parameters of one attribute and one abnormal state, and obtain its corresponding standard parameter interval [YCmin, YCmax]. Mark the operation parameter as YC0. If YC0 is within [YCmin, YCmax], it means that this attribute has no impact on this abnormal state. If YC0 is not within [YCmin, YCmax], it means that this attribute has an impact on this abnormal state. Then compare YC0 with YCmax and YCmin respectively: When YC0 > YCmax, calculate the influence difference C1 of the operation parameter of this attribute in this abnormal state through C1 = YC0 - YCmax. When YC0 < YCmin, calculate the influence difference C1 of the operation parameter of this attribute in this abnormal state through C1 = YCmin - YC0. Then, calculate the interval difference C2 of the operation parameter of this attribute in the normal state through C2 = YCmax - YCmin. After that, calculate the deviation multiple CB of the influence difference of the operation parameter of this attribute in this abnormal state through CB = C1 / C2. Similarly, the deviation multiple CB of the influence difference of all operating parameters under each abnormal state is calculated; Step 4. Select an abnormal state from each abnormal state, and obtain the deviation multiples of multiple attributes that affect the abnormal state, and then perform difference calculation and comparison processing on the deviation multiples of the multiple attributes.

[0010] Preferably, the difference calculation and comparison method in Step 4 is: Select the maximum deviation magnification and attribute of the numerical value of an operating parameter, and use them as the reference magnification value and basic attribute respectively; Then, the deviation magnifications of other attributes are subtracted from the base magnification value to obtain the corresponding magnification difference, which is then compared with the preset magnification threshold difference: If the magnification difference is greater than or equal to the magnification threshold difference, the attribute corresponding to the magnification difference is retained.

[0011] Preferably, when the magnification difference is compared with a preset magnification threshold difference: If the magnification difference is less than the magnification threshold difference, the attribute corresponding to the magnification difference is eliminated; All other attributes that are retained are then combined with the basic attributes to form the abnormal state attribute set; Next, according to the above method, multiple abnormal state attribute sets corresponding to the abnormal state are obtained, and then different attributes in the multiple abnormal state attribute sets are set to obtain all different attributes corresponding to their intersection, and then all different attributes are sorted into a standard set of abnormal state attributes; At the same time, in the abnormal state attribute standard set, the mean value of the multiplier difference corresponding to the same attribute in each abnormal state attribute set is calculated, and then the mean values ​​corresponding to each attribute are sorted from large to small, and the influence level of different attributes on it in the abnormal state is determined according to the sorting order; By analogy, multiple standard sets of abnormal state attributes are obtained.

[0012] Preferably, the standard integration process in Step 1 is as follows: First, multiple operating parameters of the same attribute obtained at multiple time nodes are marked as YCi, i=1, 2, ... n, and n represents the number of multiple operating parameters of the same attribute; Next, through , the discrete values ​​L of multiple operating parameters of the same attribute are calculated, where YCp is represented as the average value of all YCi; Then, the obtained discrete value is compared with the preset discrete threshold Ly: If L > Ly, it indicates that the dispersion degree of the operating parameters of multiple same attributes is too large. Then, the corresponding YCi values are deleted in the order from large to small of |YCi - YCp|, and the remaining deviation value L is calculated correspondingly until L ≤ Ly; After that, when L ≤ Ly, obtain the YCi values participating in the calculation of the corresponding L, and select one YCi with the largest value and one with the smallest value from them, and mark them as YCmax and YCmin respectively; Then, form the interval of the operating parameters of the corresponding attribute with YCmin and YCmax, that is, the standard parameter interval [YCmin, YCmax]; Meanwhile, when L ≤ Ly, among all the YCi values corresponding to the calculation of the corresponding L, calculate the average value of YCi, and then mark its value as the standard parameter of the corresponding attribute.

[0013] Preferably, the abnormal data extraction and processing method is as follows: During the current operation cycle of the electromechanical system, extract the real-time operation parameters of multiple different attributes at multiple same time nodes; Then, match the real-time operation parameters of different attributes at the same time node with their corresponding standard parameter intervals respectively: If the real-time operation parameters of all different attributes at the same time node are within their corresponding standard parameter intervals, it indicates that the electromechanical system is operating normally; If the real-time operation parameters of all different attributes at the same time node contain those not within their corresponding standard parameter intervals, obtain the influence difference calculated from the real-time operation parameter of the corresponding attribute and the standard parameter interval, and then calculate the deviation multiple of the influence difference of the operation parameter of this attribute in this abnormal state.

[0014] Preferably, the signal result analysis method is as follows: Rank the attributes corresponding to the deviation multiple values of each real-time operation parameter in descending order; Then, compare the ranking results with each abnormal state attribute standard set respectively for similarity: If the ranking result is similar to any one of the abnormal state attribute standard sets, it indicates that there is an abnormal state corresponding to this abnormal state attribute standard set during the current operation cycle of the electromechanical system, obtain the abnormal state corresponding to this abnormal state attribute standard set as the prediction result, and then send the display result to the data display unit; If the ranking result is not similar to any of the abnormal state attribute standard sets, it indicates that there is an unknown abnormality during the current operation cycle of the electromechanical system, and then generate an unknown abnormality signal.

[0015] Preferably, the similarity judgment criterion between the ranking result and the abnormal state attribute standard set is as follows: Compare the ranking results with each attribute corresponding to the ranking in the abnormal status attribute standard set: When the similarity rate of the same-level corresponding attributes in the results of comparing all attributes corresponding to the ranking exceeds the preset similarity rate threshold, it indicates similarity; otherwise, it indicates dissimilarity.

[0016] A method for intelligent fault prediction of complex electromechanical systems, which is implemented by the above-mentioned intelligent fault prediction system for complex electromechanical systems.

[0017] The present invention provides an intelligent fault prediction system and method for complex electromechanical systems. Compared with the prior art, it has the following beneficial effects: By analyzing the operating parameters of multiple different attributes and correlating these parameters, the present invention can identify the behavior patterns of the system in normal and abnormal states. This helps to discover in advance the factors that may cause system failures, so as to take preventive measures.

[0018] The present invention can predict potential problems before the occurrence of faults, thereby avoiding or reducing the downtime of the system. This can not only reduce production losses caused by sudden failures, but also reduce the cost of emergency repairs.

[0019] By continuously monitoring and analyzing the operating state of the system, the present invention can ensure that the system operates in the best state. This can not only extend the service life of the equipment, but also improve the overall reliability and safety of the system.

[0020] The data display unit provided by the present invention can intuitively display the prediction results and abnormal unknown signals to the management personnel, helping them make more informed decisions based on the data. This decision support is of great significance for optimizing operation strategies and improving maintenance plans.

[0021] In summary, the present invention contributes to improving the reliability of complex electromechanical systems, reducing operation and maintenance costs, and providing strong decision support for management personnel in the health management and fault prediction of complex electromechanical systems. Brief Description of the Drawings

[0022] Figure 1 is the system block diagram of the present invention; Figure 2 is the process schematic diagram of the present invention. Detailed Embodiments

[0023] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0024] Please refer to Figure 1 - Figure 2 , the present invention provides a technical solution: an intelligent fault prediction system for complex electromechanical systems, including: A data acquisition unit, configured to obtain operation parameters of multiple different attributes at multiple time nodes in the normal state and abnormal state in the early stage of the electromechanical system, and at the same time, extract real-time operation parameters of multiple different attributes at multiple different time nodes in the current operation state of the electromechanical system; Among them, the operation parameters are obtained by sensor detection, and include but are not limited to parameters such as current, temperature, humidity, voltage, etc.; An association processing unit, configured to perform association processing on the operation parameters of each different attribute; The association processing method is as follows: Step1. In the normal state in the early stage of the electromechanical system, obtain operation parameters of multiple same attributes collected from the electromechanical system at multiple time nodes, and then perform standard integration processing on the operation parameters of the same attribute to obtain standard parameters of each attribute representing the normal operation state of the electromechanical system; The standard integration processing method is as follows: First, mark the operation parameters of multiple same attributes obtained at multiple time nodes as YCi, i = 1, 2,... n, where n represents the number of operation parameters of multiple same attributes; Next, through , calculate the discrete value L of the operation parameters of multiple same attributes, where YCp represents the average value of all YCi; Then, compare the obtained discrete value with the preset discrete threshold Ly: If L > Ly, it means that the discrete degree of the operation parameters of multiple same attributes is too large. Then, delete the corresponding YCi values in the order of |YCi - YCp| from large to small and calculate the remaining deviation value L correspondingly until L ≤ Ly; After that, when L ≤ Ly, obtain the YCi participating in the calculation of the corresponding L, and select one YCi with the largest value and one YCi with the smallest value from them, and mark them as YCmax and YCmin respectively; Then, form an interval of the operation parameters of the corresponding attribute with YCmin and YCmax, that is, the standard parameter interval [YCmin, YCmax]; Meanwhile, when L ≤ Ly, calculate the average value of all YCi corresponding to the respective L during the calculation, and then mark this value as the standard parameter of the corresponding attribute; Step2. When multiple abnormal states occur in the early stage of the electromechanical system, obtain multiple operating parameters of different attributes collected from the electromechanical system at the same time node; Step3. Compare the operating parameters of each attribute in each abnormal state with their corresponding standard parameter ranges: Taking the operating parameter of one attribute and one abnormal state as an example, obtain its corresponding standard parameter range [YCmin, YCmax]; Mark this operating parameter as YC0; If YC0 is within [YCmin, YCmax], it means that this attribute has no impact on this abnormal state; If YC0 is not within [YCmin, YCmax], it means that this attribute has an impact on this abnormal state; then compare YC0 with YCmax and YCmin respectively: When YC0 > YCmax, calculate the influence difference C1 of the operating parameter of this attribute in this abnormal state through C1 = YC0 - YCmax; When YC0 < YCmin, calculate the influence difference C1 of the operating parameter of this attribute in this abnormal state through C1 = YCmin - YC0; Next, calculate the interval difference C2 of the operating parameter of this attribute in the normal state through C2 = YCmax - YCmin; After that, calculate the deviation multiple CB of the influence difference of the operating parameter of this attribute in this abnormal state through CB = C1 / C2; By analogy, calculate the deviation multiple CB of the influence difference of all operating parameters in each abnormal state; For example: Assume that in an abnormal state, the operating parameter YC0 of the corresponding temperature attribute is 35°C; In the normal state, through the analysis of the data acquisition and correlation processing unit, the standard parameter range of the temperature is obtained as [YCmin, YCmax] = [20°C, 30°C].

[0025] Step 11. Compare YC0 with the standard parameter range First, we compare YC0 with the standard parameter range. Since YC0 = 35°C exceeds the standard parameter range [20°C, 30°C], we can conclude that the temperature attribute is affected in this abnormal state.

[0026] Step 12. Calculate the influence difference C1 Since YC0 exceeds YCmax, we calculate the impact difference C1 according to the following formula: C1 = YC0 - YCmax = 35°C - 30°C = 5°C Step 13: Calculate the interval difference C2 Next, we calculate the interval difference C2 of the temperature parameters in the normal state: C2 = YCmax - YCmin = 30°C - 20°C = 10°C Step 14: Calculate the deviation magnification CB Finally, based on C1 and C2 obtained from the above calculations, we calculate the deviation magnification CB: CB = C1 / C2 = 5°C / 10°C = 0.5, and its deviation magnification CB is 0.5, indicating that the operating parameters of the temperature attribute in this abnormal state exceed the upper limit of the normal state, and the exceeding degree is half of the standard interval range; In this embodiment, the deviation magnification provides a quantitative index to facilitate understanding the specific impact degree of temperature changes on the abnormal state of the system.

[0027] Step 4: According to the comparison result of Step 3, select an abnormal state from each abnormal state and obtain the deviation magnifications of multiple attributes that affect this abnormal state; Then, calculate and compare the differences of the deviation magnifications of these multiple attributes: The method is as follows: Select the deviation magnification with the largest numerical value of an operating parameter and its attribute, and use them as the reference magnification value and the basic attribute respectively; Then subtract the deviation magnifications of other respective attributes from the reference magnification value to obtain the corresponding magnification differences, and then compare the magnification differences with the preset magnification threshold difference: If the magnification difference is greater than or equal to the magnification threshold difference, retain the attribute corresponding to the magnification difference; If the magnification difference is less than the magnification threshold difference, eliminate the attribute corresponding to the magnification difference; After that, form the abnormal state attribute set by combining all the remaining other attributes with the basic attribute; Next, according to the above method, for multiple corresponding abnormal state attribute sets belonging to the same abnormal state, then perform set operations on the different attributes in the multiple abnormal state attribute sets, and obtain all the different attributes corresponding to their intersection, and then organize these all different attributes into the abnormal state attribute standard set; Among them, if the operating parameters corresponding to multiple different attributes are in the same abnormal state attribute standard set, it means that these multiple different attributes are in an associated relationship in this abnormal state; Meanwhile, in the abnormal state attribute standard set, calculate the mean value of the ratio differences corresponding to the same attribute within each abnormal state attribute set. Then, sort the mean values corresponding to each attribute in descending order, and determine the influence level of different attributes under the abnormal state according to the sorting order. For example: Suppose in an abnormal state, the deviation ratios of three attributes, namely temperature, current, and voltage, are calculated. The deviation ratio CB value corresponding to temperature is 0.5, the deviation ratio CB value corresponding to current is 0.8, and the deviation ratio CB value corresponding to voltage is 0.3. Then, based on these data, illustrate how to generate the abnormal state attribute standard set and determine the influence levels of different attributes. Step 21: Select the reference ratio value and the basic attribute First, select the deviation ratio with the largest value as the reference ratio value. That is, the deviation ratio of current is 0.8, and the basic attribute is current.

[0028] Step 22: Calculate the ratio difference and compare it with the ratio threshold difference Set the ratio threshold difference to 0.2, and calculate the differences between the reference ratio value and the deviation ratios of other attributes: The ratio difference between current and temperature: 0.8 - 0.5 = 0.3; The ratio difference between current and voltage: 0.8 - 0.3 = 0.5; Since both of these ratio differences are greater than or equal to 0.2, retain the two attributes of temperature and voltage. Step 23: Form the abnormal state attribute set Combine the retained attributes (temperature and voltage) with the basic attribute (current) to form the abnormal state attribute set, that is, {current, temperature, voltage}.

[0029] Step 24: Generate the abnormal state attribute standard set Suppose in multiple abnormal states, multiple abnormal state attribute sets are obtained, and the intersection of these sets is obtained through set operations; For example, if the abnormal state attribute set in another abnormal state is also {current, temperature, voltage}, then the abnormal state attribute standard set is still {current, temperature, voltage}; Another example, if the abnormal state attribute set in another abnormal state is {current, temperature, humidity}, then the abnormal state attribute standard set is {current, temperature}; Step 25: Calculate the influence level In the abnormal state attribute standard set, we calculate the mean value of the ratio differences corresponding to each attribute and sort them in descending order; In this example, if the deviation multiples of current, temperature, and voltage in the mean values of all abnormal states are 0.7, 0.55, and 0.35 respectively, the sorting is: current > temperature > voltage; Based on the above analysis, we conclude that in the abnormal state, current is the property that has the greatest impact on the system, followed by temperature, and finally voltage. This sorting helps us understand the degree of influence of each property on the system in the abnormal state, so that we can prioritize monitoring and handling of the properties with greater influence to prevent or quickly respond to the abnormal state of the system.

[0030] And so on, multiple standard sets of abnormal state properties are obtained; A fault analysis unit is used to obtain real-time operating parameters of multiple different properties during the current operating cycle of the electromechanical system, and perform abnormal data extraction processing on the real-time operating parameters through an association processing unit to obtain corresponding deviation multiples, and based on the respective properties corresponding to the deviation multiples, perform signal result analysis on each standard set of abnormal state properties and obtain corresponding prediction results or abnormal unknown signals; The specific method of abnormal data extraction processing is as follows: During the current operating cycle of the electromechanical system, extract real-time operating parameters of multiple different properties at multiple same time nodes; Then match the real-time operating parameters of different properties at the same time node with their corresponding standard parameter intervals respectively: If the real-time operating parameters of all different properties at the same time node are within their corresponding standard parameter intervals, it means that the electromechanical system is operating normally; If the real-time operating parameters of all different properties at the same time node contain those that are not within their corresponding standard parameter intervals, obtain the influence difference calculated from the real-time operating parameters of the corresponding property and the standard parameter interval, and then calculate the deviation multiple of the influence difference of the operating parameters of this property in this abnormal state; The specific method of signal result analysis is as follows: Then, sort the properties corresponding to each deviation multiple value in descending order; That is, in descending order, take the group with the largest deviation multiple value of the corresponding property as sorting rank one, then take the group with the second largest deviation multiple value of the corresponding property as sorting rank two, and so on..., and obtain the corresponding rank sorting result; Then compare the rank sorting result with each standard set of abnormal state properties for similarity: The similarity comparison between the rank sorting result and the standard set of abnormal state properties is as follows: Compare the rank sorting result with each property corresponding to the rank sorting in the standard set of abnormal state properties: When, among the results of comparing all properties corresponding to the hierarchical sorting, the similarity rate of the same-level corresponding properties exceeds the preset similarity rate threshold, it indicates similarity; otherwise, it indicates dissimilarity. According to the similarity comparison result: If the hierarchical sorting result is similar to any one of the abnormal state property standard sets, it indicates that there is an abnormal state corresponding to the abnormal state property standard set during the current operation cycle of the electromechanical system, and the abnormal state corresponding to the abnormal state property standard set is obtained as the prediction result, and then the display result is sent to the data display unit. If the hierarchical sorting result is not similar to any of the abnormal state property standard sets, it indicates that there is an unknown abnormality during the current operation cycle of the electromechanical system, and then an abnormal unknown signal is generated. For example: If in the hierarchical sorting result, the corresponding hierarchical sortings are respectively Property 2, Property 4, Property 6, Property 8, Property 1, Property 9, Property 7, Property 12, Property 13 in sequence; If in the abnormal state property standard set, the corresponding hierarchical sortings are respectively Property 2, Property 4, Property 5, Property 8, Property 1, Property 9, Property 7, Property 12 in sequence; Then the property of the first hierarchical sorting is the same, the property of the second hierarchical sorting is the same, the property of the third hierarchical sorting is different, the property of the fourth hierarchical sorting is the same, the property of the fifth hierarchical sorting is the same, the property of the sixth hierarchical sorting is the same, the property of the seventh hierarchical sorting is the same, the property of the eighth hierarchical sorting is the same, the property of the ninth hierarchical sorting is not included in the abnormal state property standard set, so it does not participate in the calculation of the similarity rate, and its similarity rate is 0.875; If the similarity rate threshold is 0.85 and 0.875 is greater than 0.85, it indicates that the hierarchical sorting result is similar to the abnormal state property standard set; If in the abnormal state property standard set, the corresponding hierarchical sortings are respectively Property 2, Property 4, Property 5, Property 8, Property 6, Property 9, Property 7, Property 12 in sequence; Then the property of the first hierarchical sorting is the same, the property of the second hierarchical sorting is the same, the property of the third hierarchical sorting is different, the property of the fourth hierarchical sorting is the same, the property of the fifth hierarchical sorting is the same, the property of the sixth hierarchical sorting is the same, the property of the seventh hierarchical sorting is the same, the property of the eighth hierarchical sorting is the same, the property of the ninth hierarchical sorting is not included in the abnormal state property standard set, so it does not participate in the calculation of the similarity rate, and its similarity rate is 0.875; If the similarity rate threshold is 0.85 and 0.875 is greater than 0.85, it indicates that the hierarchical sorting result is similar to the abnormal state property standard set; Then the attributes sorted in the first level are the same, the attributes sorted in the second level are the same, the attributes sorted in the third level are different, the attributes sorted in the fourth level are the same, the attributes sorted in the fifth level are different, the attributes sorted in the sixth level are the same, the attributes sorted in the seventh level are the same, the attributes sorted in the eighth level are the same, and the attributes sorted in the ninth level are not included in the abnormal state attribute standard set, so it does not participate in the calculation of the similarity rate, and its similarity rate is 0.75; If the similarity rate threshold is 0.85 and 0.75 is less than 0.85, it means that the ranking result is not similar to the abnormal state attribute standard set; This embodiment provides a specific similarity judgment criterion. By calculating the similarity rate of the attributes corresponding to the same level in the ranking result and the abnormal state attribute standard set and comparing it with the preset similarity rate threshold, it is determined whether they are similar.

[0031] This method simplifies the similarity judgment process, making the fault prediction system more efficient and easy to implement; by setting the similarity rate threshold, the system can flexibly adjust the prediction sensitivity to meet the requirements of different application scenarios; Among them, if the data received by the data display unit is a prediction result, it means that the fault information existing in the electromechanical system is a known fault, and the management personnel can perform maintenance processing through the previous maintenance data; Among them, if the data received by the data display unit is an abnormal unknown signal, it means that the fault information existing in the electromechanical system is an unknown fault, and the management personnel need to promptly conduct fault investigation and maintenance on the electromechanical system according to the operating parameters of its corresponding attributes; The data display unit is used to display the prediction result and the abnormal unknown signal corresponding to the current operation cycle of the electromechanical system to the management personnel; In this embodiment, by collecting the operating parameters of different attributes in the normal state and abnormal state of the electromechanical system in the early stage, this solution can establish a standard parameter range to provide a benchmark for subsequent fault prediction.

[0032] This embodiment uses the association processing unit to perform standard integration processing and comparative analysis on the operating parameters, which can effectively identify which attributes have an impact on the abnormal state of the electromechanical system and calculate the impact difference and deviation multiple.

[0033] The fault analysis unit can extract the real-time operating parameters during the current operation cycle of the electromechanical system and match them with the standard parameter range to promptly detect the abnormal state and improve the fault prediction efficiency.

[0034] The data display unit intuitively displays the prediction result and the abnormal unknown signal to the management personnel to help them quickly understand the system state and take corresponding measures.

[0035] The entire system realizes the intelligent prediction of mechanical and electrical system failures, reduces the cost of manual inspection, and improves the operation and maintenance efficiency and equipment reliability.

[0036] The present invention also provides a technical solution, a method for intelligent prediction of complex mechanical and electrical system failures. This method is implemented according to a complex mechanical and electrical system failure intelligent prediction system, and includes the following steps: First step, data collection Under the normal state and abnormal state in the early stage of the mechanical and electrical system, obtain the operating parameters of multiple different attributes at multiple time nodes. At the same time, in the current operating state of the mechanical and electrical system, extract the real-time operating parameters of multiple different attributes at multiple different time nodes. Among them, the operating parameters are obtained by sensor detection and include but are not limited to parameters such as current, temperature, humidity, voltage, etc.

[0037] Second step, data association Perform association processing on the operating parameters of each different attribute. First, when the mechanical and electrical system is in the normal state in the early stage, obtain the operating parameters of multiple same attributes collected from the mechanical and electrical system at multiple time nodes, and then perform standard integration processing on the operating parameters of the same attribute to obtain the standard parameters of each attribute representing the normal operating state of the mechanical and electrical system. Then, when the mechanical and electrical system has multiple abnormal states in the early stage, obtain the operating parameters of multiple different attributes collected from the mechanical and electrical system at the same time node. Next, compare the operating parameters of each attribute in each abnormal state with the corresponding standard parameter interval. Finally, according to the comparison results, select an abnormal state from each abnormal state and obtain the deviation multiples of multiple attributes that affect this abnormal state.

[0038] Third step, data analysis Extract abnormal data from the real-time operating parameters of multiple different attributes obtained during the current operating cycle of the mechanical and electrical system and obtain the corresponding deviation multiples. According to the attributes corresponding to the deviation multiples, perform signal result analysis on each abnormal state attribute standard set and obtain the corresponding prediction results or abnormal unknown signals.

[0039] Fourth step, data display Display the corresponding prediction results and abnormal unknown signals during the current operating cycle of the mechanical and electrical system to the management personnel. If the received data is a prediction result, it means that the fault information existing in the mechanical and electrical system is a known fault, and the management personnel can perform maintenance processing based on the previous maintenance data; if the received data is an abnormal unknown signal, it means that the fault information existing in the mechanical and electrical system is an unknown fault, and it is necessary for the management personnel to promptly perform fault troubleshooting and maintenance on the mechanical and electrical system according to the operating parameters of its corresponding attributes.

[0040] This comprehensive method makes the system more comprehensive and powerful, capable of meeting the more complex fault prediction requirements of electromechanical systems.

[0041] Generally speaking, the intelligent fault prediction system and method for complex electromechanical systems provided by these embodiments can effectively improve the maintenance efficiency of electromechanical systems, reduce the failure rate, ensure the stable operation of the system, and provide strong decision-making support for managers.

[0042] The above formulas are all dimensionless and take their numerical calculations. The formulas are obtained by collecting a large amount of data for software simulation to get a formula closest to the actual situation. The preset parameters and threshold selection in the formulas are set by those skilled in the art according to the actual situation.

[0043] Meanwhile, the content not described in detail in this specification belongs to the prior art well-known to those skilled in the art.

[0044] The above has described in detail one embodiment of the present invention, but the content is only a preferred embodiment of the present invention and cannot be considered as limiting the scope of implementation of the present invention. Any equivalent changes and improvements made within the scope of the application of the present invention shall still fall within the scope covered by the patent of the present invention.

Claims

1. An intelligent fault prediction system for complex electromechanical systems, characterized in that: include: Data acquisition unit, used to obtain multiple operating parameters of different attributes in the electromechanical system, An association processing unit is used to associate the operating parameters of different attributes to obtain multiple abnormal state attribute standard sets; wherein the abnormal state attribute standard set includes operating parameters corresponding to multiple different attributes, and they are associated in the abnormal state; The fault analysis unit is used to obtain real-time operating parameters of multiple different attributes in the current operating cycle of the electromechanical system, perform abnormal data extraction and processing, and obtain the corresponding deviation ratio, and according to the various attributes corresponding to the deviation ratio, perform signal result analysis on each abnormal state attribute standard set and obtain the corresponding prediction result or abnormal unknown signal; The data display unit is used to display the corresponding prediction results and abnormal unknown signals in the current operation cycle of the electromechanical system to the management personnel.

2. The complex electromechanical system fault intelligent prediction system according to claim 1, characterized in that: The operating parameters are obtained through sensor monitoring, and include but are not limited to current parameters, temperature parameters, humidity parameters, and voltage parameters.

3. The complex electromechanical system fault intelligent prediction system according to claim 1 is characterized by: The association is handled as follows: Step 1, when the electromechanical system is in a normal state in the early stage, multiple operating parameters of the same attribute collected from the electromechanical system at multiple time nodes are obtained, and then the operating parameters of the same attribute are subjected to standard integration processing, and standard parameters of various attributes representing the normal operating state of the electromechanical system are obtained; Step 2, when the electromechanical system has multiple abnormal states in the early stage, obtain multiple operating parameters with different attributes collected from the electromechanical system at the same time node; Step 3. Compare the operating parameters of each attribute under each abnormal state with the corresponding standard parameter range: Select an operating parameter of an attribute and an abnormal state, and obtain its corresponding standard parameter range [YCmin, YCmax]; This operating parameter is marked as YC0; If YC0 is within [YCmin, YCmax], it means that the attribute has no impact on the abnormal state; If YC0 is not within [YCmin, YCmax], it means that the attribute affects the abnormal state; then YC0 is compared with YCmax and YCmin respectively: When YC0>YCmax, the influence difference C1 of the running parameters of the attribute under the abnormal state is calculated through C1=YC0-YCmax; When YC0<YCmin, the influence difference C1 of the operating parameters of the attribute under the abnormal state is calculated by C1=YCmin-YC0; Next, the interval difference C2 of the operating parameters of the attribute under normal conditions is calculated through C2=YCmax-YCmin; Then, the deviation multiple CB of the influence difference of the operating parameters of the attribute under the abnormal state is calculated by CB=C1 / C2; Similarly, the deviation multiple CB of the influence difference of all operating parameters under each abnormal state is calculated; Step 4. Select an abnormal state from each abnormal state, and obtain the deviation multiples of multiple attributes that affect the abnormal state, and then perform difference calculation and comparison processing on the deviation multiples of the multiple attributes.

4. The complex electromechanical system fault intelligent prediction system according to claim 3 is characterized by: The difference calculation and comparison method in Step 4 is: Select the maximum deviation magnification and attribute of the numerical value of an operating parameter, and use them as the reference magnification value and basic attribute respectively; Then, the deviation magnifications of other attributes are subtracted from the base magnification value to obtain the corresponding magnification difference, which is then compared with the preset magnification threshold difference: If the magnification difference is greater than or equal to the magnification threshold difference, the attribute corresponding to the magnification difference is retained.

5. The complex electromechanical system fault intelligent prediction system according to claim 4 is characterized by: When the magnification difference is compared with the preset magnification threshold difference: If the magnification difference is less than the magnification threshold difference, the attribute corresponding to the magnification difference is eliminated; All other attributes that are retained are then combined with the basic attributes to form the abnormal state attribute set; Next, according to the above method, multiple abnormal state attribute sets corresponding to the abnormal state are obtained, and then different attributes in the multiple abnormal state attribute sets are set to obtain all different attributes corresponding to their intersection, and then all different attributes are sorted into a standard set of abnormal state attributes; At the same time, in the abnormal state attribute standard set, the mean value of the multiplier difference corresponding to the same attribute in each abnormal state attribute set is calculated, and then the mean values ​​corresponding to each attribute are sorted from large to small, and the influence level of different attributes on it in the abnormal state is determined according to the sorting order; By analogy, multiple standard sets of abnormal state attributes are obtained.

6. The complex electromechanical system fault intelligent prediction system according to claim 3 is characterized by: The standard integration process in Step 1 is as follows: First, multiple operating parameters of the same attribute obtained at multiple time nodes are marked as YCi, i=1, 2, ... n, and n represents the number of multiple operating parameters of the same attribute; Next, through , the discrete values ​​L of multiple operating parameters of the same attribute are calculated, where YCp is represented as the average value of all YCi; Then, the obtained discrete value is compared with the preset discrete threshold Ly: If L>Ly, it means that the discrete degree of multiple operating parameters of the same attribute is too large. Then, the corresponding YCi values ​​are deleted in the order of |YCi-YCp| from large to small, and the remaining deviation value L is calculated accordingly until L≤Ly; When L≤Ly is obtained, YCi of the corresponding L is calculated, and a YCi with the maximum value and a YCi with the minimum value are selected and marked as YCmax and YCmin respectively; Then YCmin and YCmax are combined into the interval of the operating parameters of the corresponding attributes, that is, the standard parameter interval [YCmin, YCmax]; At the same time, when L≤Ly, participate in the calculation of all YCi corresponding to the corresponding L, find the average value of YCi, and then mark its value as the standard parameter of the corresponding attribute.

7. The complex electromechanical system fault intelligent prediction system according to claim 3 is characterized by: Abnormal data extraction and processing are as follows: Extracting real-time operating parameters of multiple different attributes at multiple same time nodes during the current operating cycle of the electromechanical system; Then, the real-time running parameters of different attributes at the same time node are matched with their corresponding standard parameter ranges: If all real-time operating parameters of different attributes at the same time node are within their corresponding standard parameter ranges, it means that the electromechanical system is operating normally; If the real-time operating parameters of all different attributes of the same time node are not in their corresponding standard parameter intervals, obtain the impact difference calculated between the real-time operating parameters of the corresponding attribute and the standard parameter interval, and then calculate the deviation ratio of the impact difference of the operating parameters of the attribute under this abnormal state according to Step 3.

8. The complex electromechanical system fault intelligent prediction system according to claim 7 is characterized by: The signal result analysis method is as follows: According to the order from large to small, the attributes corresponding to the deviation magnification values ​​corresponding to the real-time operating parameters are ranked; Then the ranking results are compared with the standard sets of each abnormal state attribute for similarity: If the ranking result is similar to any of the abnormal state attribute standard sets, it means that an abnormal state corresponding to the abnormal state attribute standard set exists in the current operation cycle of the electromechanical system, and the abnormal state corresponding to the abnormal state attribute standard set is obtained as the prediction result, and the display result is then sent to the data display unit; If the ranking result is not similar to any of the standard sets of abnormal state attributes, it means that there are unknown abnormalities in the current operation cycle of the electromechanical system, and an abnormal unknown signal is generated.

9. The complex electromechanical system fault intelligent prediction system according to claim 8, characterized in that: in, The similarity judgment criteria between the ranking results and the abnormal state attribute standard set are as follows: Compare the ranking results with the attributes in the abnormal state attribute standard set, and the corresponding ranking attributes: When the same rate of the corresponding attributes of the same level in the comparison results of all the attributes sorted by the corresponding levels exceeds the preset same rate threshold, it means that they are similar, otherwise, it means that they are not similar.

10. A complex electromechanical system fault intelligent prediction method, characterized in that: The method is implemented by a complex electromechanical system fault intelligent prediction system as described in any one of claims 1-9.

Images