AI-Based Abnormal Prediction System and Method for Floor Scrubbing Machine Production Equipment

Abstract

The present application provides an AI-based abnormal prediction system and method for floor scrubber production equipment. Among them, by monitoring the operating parameters of the floor scrubber production equipment in real time and comparing them with historical data, the periodic and non-periodic components are separated; a clustering algorithm is used to identify unexpected fluctuation patterns, and key factors are obtained through association rule mining; a benchmark performance curve is constructed by combining the performance data of the same type of components, and based on the degree of deviation of the actual performance curve, a fuzzy logic system is used to evaluate the potential abnormal risk level; a Bayesian network is used to predict possible failures within a specific future time period, and a maintenance recommendation report is generated. The present application improves the maintenance efficiency and reliability of the floor scrubber production equipment.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of AI technology, and in particular to an abnormal prediction system and method for floor scrubber production equipment based on AI. Background Art

[0002] In a modern industrial production environment, the stable operation of floor scrubber production equipment is crucial for the overall efficiency and product quality of the production line. To ensure that the equipment can maintain efficient operation for a long time, it is particularly critical to monitor its operating parameters (such as temperature, pressure, vibration, etc.) in real time and compare them with historical data. Through this comparison, the periodic and non-periodic components in the operating parameters can be separated, so as to identify unexpected fluctuation patterns that may affect the normal operation of the equipment. Further using clustering algorithms and association rule mining algorithms, not only can the causal relationships and co-occurrence patterns behind these fluctuation patterns be discovered, but also key factors can be extracted to provide a basis for subsequent equipment health assessment. Combining the performance data of the same type of components to construct a benchmark performance curve, and according to the deviation degree between the actual performance curve and the benchmark performance curve, using a fuzzy logic system to evaluate the potential abnormal risk level. Finally, a Bayesian network is used to predict the equipment that may fail in a specific future time period and generate a maintenance recommendation report to achieve preventive maintenance, reduce unexpected downtime, extend the service life of the equipment, and improve the overall production efficiency.

[0003] Currently, the monitoring and maintenance of floor scrubber production equipment mainly rely on traditional threshold alarm mechanisms and time series analysis methods based on statistics. The traditional threshold alarm mechanism monitors the changes of single or multiple operating parameters by setting fixed upper and lower limits, and once a certain parameter exceeds the preset range, an alarm is triggered. In addition, some enterprises have adopted time series analysis methods based on statistics to try to identify abnormal behaviors of the equipment from historical data. However, most of these methods can only process single-dimensional data and lack a comprehensive understanding of the complex relationships between multiple parameters. For example, although some solutions can detect abnormal changes in certain parameters, they cannot accurately judge whether these changes really indicate the occurrence of equipment failures. At the same time, the existing solutions are less efficient in processing large-scale data sets and are difficult to achieve real-time monitoring and rapid response, which limits their application effects in a dynamically changing environment.

[0004] Although existing monitoring and maintenance solutions meet the basic requirements to a certain extent, they have obvious limitations. First, traditional solutions focus on single-point monitoring and static threshold settings, making it difficult to adapt to dynamic changes under different working conditions, and prone to false alarms or missed alarms, which affect the accuracy of maintenance decisions. Second, existing methods lack the ability to effectively predict future failures and cannot formulate preventive maintenance plans in advance, increasing the risk of sudden equipment failures. In addition, these solutions perform poorly in processing complex multi-dimensional data and are difficult to extract valuable information from massive data, limiting their application in high-precision equipment health management. Summary of the Invention

[0005] The embodiments of the present application provide an AI-based abnormal prediction system and method for floor scrubber production equipment to solve the problems of low maintenance efficiency and poor reliability of floor scrubber production equipment in the prior art.

[0006] In a first aspect, the embodiments of the present application provide an AI-based abnormal prediction method for floor scrubber production equipment, including:

[0007] Receiving operation parameters from real-time monitored floor scrubber production equipment, comparing the operation parameters with historical data from the operation of the floor scrubber production equipment, and obtaining periodic and non-periodic components in the operation parameters;

[0008] According to the periodic and non-periodic components, using a clustering algorithm to identify unexpected fluctuation patterns in the operation parameters;

[0009] Using an association rule mining algorithm to perform causal relationship and co-occurrence pattern discovery processing on the unexpected fluctuation patterns to obtain key factors;

[0010] Using the key factors as inputs, combining performance data of the same type of components among floor scrubber production equipment to construct a benchmark performance curve, and based on the deviation degree between the benchmark performance curve and the actual performance curve, combining a fuzzy logic system to evaluate the floor scrubber production equipment and determine the potential abnormal risk level;

[0011] According to the unexpected fluctuation patterns, the key factors, and the potential abnormal risk level, using a Bayesian network to predict floor scrubber production equipment that will fail within a preset time period and generating a maintenance recommendation report.

[0012] Optionally, the step of using the key factors as inputs, combining performance data of the same type of components among floor scrubber production equipment to construct a benchmark performance curve, and based on the deviation degree between the benchmark performance curve and the actual performance curve, combining a fuzzy logic system to evaluate the floor scrubber production equipment and determine the potential abnormal risk level includes:

[0013] Using the key factors as inputs, combining the performance data of the same type of components among the floor scrubber production equipment, applying an adaptive filtering algorithm to smooth the multi-source historical performance data, obtaining the smoothed data, using a support vector regression model to fit the smoothed data, generating a benchmark performance curve, where the multi-source historical performance data includes: the performance data of the same type of components, the actual operation parameters of the current equipment at different time periods, historical failure records, and maintenance logs;

[0014] By comparing the actual performance curve with the benchmark performance curve, calculating the deviation degree between the actual performance curve and the benchmark performance curve, and quantifying the deviation degree using the root mean square error, obtaining a deviation degree score;

[0015] According to the deviation degree score, combining with a fuzzy logic system to evaluate the floor scrubber production equipment, obtaining an output result, and using a particle swarm optimization algorithm to optimize the output result, obtaining the health status evaluation result of the floor scrubber production equipment;

[0016] Using the health status evaluation result, combining with a time series prediction model, predicting the performance change trend of the floor scrubber production equipment within a preset time period, obtaining a prediction result, and analyzing the prediction result to determine the potential abnormal risk level.

[0017] Optionally, the step of according to the deviation degree score, combining with a fuzzy logic system to evaluate the floor scrubber production equipment, obtaining an output result, and using a particle swarm optimization algorithm to optimize the output result, obtaining the health status evaluation result of the floor scrubber production equipment includes:

[0018] Using an adaptive neuro-fuzzy inference system to perform fuzzy logic evaluation processing on the deviation degree score, generating a preliminary evaluation result;

[0019] Based on the preliminary evaluation result, combining the genetic algorithm and the differential evolution algorithm to optimize the fuzzy logic system, obtaining an optimized fuzzy logic system;

[0020] Inputting the deviation degree score into the optimized fuzzy logic system, generating an output result, and using a particle swarm optimization algorithm and a simulated annealing algorithm to optimize the output result, generating an optimized output result;

[0021] According to the ensemble learning method, comprehensively combining the preliminary evaluation result, the optimized fuzzy logic system, and the optimized output result, generating the health status evaluation result of the floor scrubber production equipment.

[0022] Optionally, inputting the deviation degree score into the optimized fuzzy logic system to generate an output result, and using the particle swarm optimization algorithm and the simulated annealing algorithm to optimize the output result to generate an optimized output result, including:

[0023] Using the optimized fuzzy logic system, taking the deviation degree score as input data, performing fuzzy inference processing on the deviation degree score, and outputting a preliminary evaluation result of the equipment health status;

[0024] Using the particle swarm optimization algorithm to perform a global search on the preliminary evaluation result of the equipment health status, identifying the optimal solution, and optimizing the preliminary evaluation result of the equipment health status based on the optimal solution to obtain the result of the particle swarm optimization algorithm;

[0025] According to the result of the particle swarm optimization algorithm, combined with the simulated annealing algorithm, optimizing the preliminary evaluation result of the equipment health status again to generate an optimized output result.

[0026] Optionally, using the association rule mining algorithm to perform causal relationship and co-occurrence pattern discovery processing on the unexpected fluctuation pattern to obtain key factors, including:

[0027] Using the unexpected fluctuation pattern to construct a data set, performing dimensionality reduction processing on the data set using principal component analysis to obtain an optimized data set, and using the association rule mining algorithm to scan the optimized data set to identify frequent item sets;

[0028] Using Granger causality test to detect the frequent item sets to obtain the causal relationship of the operating parameters, and extracting the co-occurrence pattern from the frequent item sets, and combining the causal relationship and the co-occurrence pattern to obtain a key set;

[0029] Using lift, Jaccard similarity coefficient, and Sørensen-Dice coefficient as evaluation indicators to evaluate the causal relationship and co-occurrence pattern of the key set to generate a comprehensive scoring result;

[0030] According to the comprehensive scoring result, screening out candidate key factors from the frequent item sets, and using the random forest model to perform importance scoring processing on the candidate key factors to obtain key factors.

[0031] Optionally, using lift, Jaccard similarity coefficient, and Sørensen-Dice coefficient as evaluation indicators to evaluate the causal relationship and co-occurrence pattern of the key set to generate a comprehensive scoring result, including:

[0032] Calculating the lift of the causal relationship and co-occurrence pattern in the key set to obtain a correlation strength score;

[0033] Using the Jaccard similarity coefficient, perform similarity evaluation processing on the causal relationships and co-occurrence patterns in the key set, calculate the similarity degree of the causal relationships and co-occurrence patterns in the key set, and obtain a similarity score;

[0034] Using the Sørensen-Dice coefficient, perform uniqueness and importance evaluation processing on the causal relationships and the co-occurrence patterns, calculate the uniqueness scores of the causal relationships and the co-occurrence patterns, and obtain a uniqueness score;

[0035] Combining the correlation strength score, the similarity score, and the uniqueness score, perform comprehensive scoring processing on each causal relationship and co-occurrence pattern in the key set, and obtain a comprehensive scoring result.

[0036] Optionally, the identifying the unexpected fluctuation patterns occurring in the operating parameters by using a clustering algorithm according to the periodic and aperiodic components includes:

[0037] Based on the periodic and aperiodic components, use a clustering algorithm to classify the operating parameters, identify multiple operating modes and abnormal modes, and obtain a clustering result according to the operating modes and the abnormal modes;

[0038] By comparing the clustering result with the known normal operating modes, identify and mark the operating parameters that do not conform to the normal operating modes as unexpected fluctuation patterns.

[0039] In a second aspect, an abnormal prediction system for a floor mopping machine production device based on AI provided by an embodiment of the present application includes:

[0040] A receiving module, configured to receive the operating parameters of the floor mopping machine production device from real-time monitoring, compare the operating parameters with the historical data of the operation of the floor mopping machine production device, and obtain the periodic and aperiodic components in the operating parameters;

[0041] An identifying module, configured to identify the unexpected fluctuation patterns occurring in the operating parameters by using a clustering algorithm according to the periodic and aperiodic components;

[0042] A utilization module, configured to perform causal relationship and co-occurrence pattern discovery processing on the unexpected fluctuation patterns by using an association rule mining algorithm to obtain key factors;

[0043] A construction module, configured to use the key factors as inputs, combine the performance data of the same type of components among the floor mopping machine production devices to construct a benchmark performance curve, and based on the deviation degree between the benchmark performance curve and the actual performance curve, evaluate the floor mopping machine production device by combining a fuzzy logic system to determine the potential abnormal risk level;

[0044] A prediction module, configured to predict, according to the unexpected fluctuation pattern, the key factors, and the potential abnormal risk level, the floor scrubber production equipment that may fail within a preset time period by using a Bayesian network, and generate a maintenance recommendation report.

[0045] In a third aspect, an embodiment of the present application provides a computing device, including a processor and a memory. A computer program is stored in the memory, and the processor is configured to run the computer program to execute the AI-based abnormal prediction method for floor scrubber production equipment according to any one of the first aspect.

[0046] In a fourth aspect, an embodiment of the present application provides a computer storage medium, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the AI-based abnormal prediction method for floor scrubber production equipment according to any one of the first aspect is implemented.

[0047] In the embodiment of the present application, operation parameters from real-time monitored floor scrubber production equipment are received, the operation parameters are compared with historical data from the operation of the floor scrubber production equipment to obtain the periodic and non-periodic components in the operation parameters; according to the periodic and non-periodic components, a clustering algorithm is used to identify the unexpected fluctuation patterns in the operation parameters; an association rule mining algorithm is used to perform causal relationship and co-occurrence pattern discovery processing on the unexpected fluctuation patterns to obtain key factors; the key factors are used as inputs, combined with the performance data of the same type of components between floor scrubber production equipment, a benchmark performance curve is constructed, and based on the deviation degree between the benchmark performance curve and the actual performance curve, a floor scrubber production equipment is evaluated by combining a fuzzy logic system to determine the potential abnormal risk level; according to the unexpected fluctuation pattern, the key factors, and the potential abnormal risk level, a Bayesian network is used to predict the floor scrubber production equipment that may fail within a preset time period, and a maintenance recommendation report is generated.

[0048] The technical solution of the present application has the following beneficial effects:

[0049] By comprehensively applying various advanced algorithms such as cluster analysis, association rule mining, fuzzy logic system, and Bayesian network, the present application can not only accurately identify the unexpected fluctuation patterns and their causal relationships in equipment operation, but also accurately evaluate the potential abnormal risk level and predict the future failure probability. This method significantly improves the predictability and effectiveness of equipment maintenance, reduces unexpected downtime, extends the service life of equipment, and greatly improves production efficiency and economic benefits.

[0050] Furthermore, in the embodiments of the present application, key factors are also used as inputs, combined with the performance data of the same type of components among the floor scrubber production equipment, and the adaptive filtering algorithm is applied to smooth the multi-source historical performance data to obtain the smoothed data. Then, the support vector regression model is used to fit these data to generate a benchmark performance curve. By comparing the actual performance curve with the benchmark performance curve, the deviation degree is calculated and quantified using the root mean square error to obtain a deviation degree score. Based on the deviation degree score, the fuzzy logic system is used to evaluate the equipment health status, and the particle swarm optimization algorithm is used to optimize the output result, and finally the equipment health status evaluation result is generated. In addition, the time series prediction model is combined to predict the equipment performance change trend within a preset time period, and the prediction results are analyzed to determine the potential abnormal risk level.

[0051] Through the above method, not only can the benchmark performance curve be accurately constructed and the deviation degree between the actual performance curve and the benchmark performance curve be quantified, but also the health status of the floor scrubber production equipment can be effectively evaluated. Specifically, by introducing key factors and multi-source historical performance data (such as the performance data of the same type of components, actual operation parameters, historical failure records, and maintenance logs), the accuracy and reliability of the benchmark performance curve are improved. Further combined with the fuzzy logic system and the particle swarm optimization algorithm, the accurate evaluation and optimization processing of the equipment health status are realized. Finally, the time series prediction model is used to predict the equipment performance change trend within a specific future time period, which helps to identify potential abnormal risks in advance, formulate preventive maintenance strategies, thereby significantly improving the operation efficiency and reliability of the equipment, reducing unexpected downtime and maintenance costs, and improving the overall production efficiency. This method has stronger predictability and higher accuracy compared with traditional solutions and is applicable to complex and changeable industrial environments.

[0052] These aspects or other aspects of the present application will be more clearly understood in the following description of the embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0054] Figure 1 It is a flowchart of the method for predicting anomalies in floor scrubber production equipment based on AI provided by the embodiments of the present application;

[0055] Figure 2 It is a schematic structural diagram of the system for predicting anomalies in floor scrubber production equipment based on AI provided by the embodiments of the present application;

[0056] Figure 3 A structural schematic diagram of a computing device provided by an embodiment of the present application. Detailed implementation manners

[0057] In order to enable those skilled in the art of the present technology to better understand the solutions of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present application.

[0058] In some processes described in the specification, claims and the above-mentioned accompanying drawings of the present application, a plurality of operations that appear in a specific order are included. However, it should be clearly understood that these operations may not be executed in the order in which they appear in this article or may be executed in parallel. The serial numbers of the operations, such as 101, 102, etc., are only used to distinguish each different operation, and the serial numbers themselves do not represent any execution order. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions such as "first" and "second" in this article are used to distinguish different messages, devices, modules, etc., do not represent the order of precedence, and do not limit that "first" and "second" are of different types.

[0059] The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative efforts belong to the scope of protection of the present application.

[0060] Figure 1 A flowchart of an abnormal prediction method for a floor scrubber production device based on AI provided by an embodiment of the present application is as Figure 1 shown, and the method includes:

[0061] Step 101: Receive the operation parameters of the floor scrubber production device from real-time monitoring, compare the operation parameters with the historical data of the operation of the floor scrubber production device, and obtain the periodic and non-periodic components in the operation parameters;

[0062] In this step, the operation parameters refer to various data generated in real time during the operation of the device, such as temperature, pressure, vibration, etc. These parameters reflect the current working state of the device. The historical data is the operation records of the same or similar devices in the past period of time. By comparing the current operation parameters with the historical data, the periodic component (reflecting the normal working mode) and the non-periodic component (which may indicate abnormal behavior or potential faults) can be separated.

[0063] In actual operation, first, the operating parameters of the floor scrubber production equipment are collected in real time through the sensor network. Then, using signal processing techniques, these parameters are decomposed into periodic and aperiodic components. The periodic components usually represent the normal operating mode of the equipment, while the aperiodic components may indicate potential problems.

[0064] For example, in a large manufacturing plant, engineers installed high-precision sensors to monitor the operation of multiple floor scrubbers. They collected the temperature and vibration data of each device and conducted a comparative analysis with the historical data of the past year. It was found that the vibration levels of certain devices increased significantly during specific time periods, which may be abnormal behavior caused by mechanical component wear.

[0065] Step 102: According to the periodic and aperiodic components, use the clustering algorithm to identify the unexpected fluctuation patterns that appear in the operating parameters;

[0066] In this step, the clustering algorithm is an unsupervised learning method used to divide the objects in a dataset into multiple groups or clusters, such that the objects within the same group are more similar to each other than those in other groups. In this context, the clustering algorithm is used to identify the unexpected fluctuation patterns in the operating parameters, that is, those data points or patterns that deviate from the normal operating mode.

[0067] In actual operation, select an appropriate clustering algorithm and use the periodic and aperiodic components obtained in step 101 as feature vectors and input them into the algorithm. The larger clusters in the clustering results usually represent the normal mode, while the smaller clusters or isolated points may be abnormal modes.

[0068] For example, continuing the above case, after identifying the periodic and aperiodic components, the engineers applied the clustering algorithm to further analyze the data. The results showed that significant unexpected fluctuations occurred in certain devices during specific time periods, and these fluctuations may indicate impending failures. This information provided important clues for further analysis.

[0069] Step 103: Use the association rule mining algorithm to perform causal relationship and co-occurrence pattern discovery processing on the unexpected fluctuation patterns to obtain key factors;

[0070] In this step, association rule mining is a data mining technique used to discover interesting relationships between variables in a database. In this scenario, it is used to analyze the causal relationships and co-occurrence patterns in the unexpected fluctuation patterns to extract key factors. These key factors can help determine which conditions or events may cause abnormal behavior of the device.

[0071] In actual operation, through the Apriori algorithm or other association rule mining methods, analyze the mutual relationships among various variables in the unexpected fluctuation patterns to determine which factors co-occur or cause anomalies. The finally generated key factors can be used for constructing the benchmark performance curve in the subsequent steps.

[0072] For example, based on the previous step, engineers used the Apriori algorithm to conduct an in-depth analysis of the unexpected fluctuation patterns. The results showed that when certain devices operated for a long time in a high-temperature environment, the vibration level increased significantly, which might be due to the failure of the cooling system. These key factors were recorded and used for constructing the benchmark performance curve in the next step.

[0073] Step 104: Use the key factors as inputs, combine with the performance data of the same type of components among the floor scrubber production equipment, construct a benchmark performance curve, and based on the deviation degree between the benchmark performance curve and the actual performance curve, evaluate the floor scrubber production equipment in combination with a fuzzy logic system to determine the potential abnormal risk level;

[0074] In this step, the adaptive filtering algorithm is a technology that dynamically adjusts its parameters to adapt to the changing environment and is commonly used in the field of signal processing. Support vector regression is an extended form of support vector machines and is used for regression analysis. The benchmark performance curve is an ideal performance reference line constructed based on historical data and current key factors and is used to evaluate the deviation degree of the actual equipment performance.

[0075] In actual operation, first, apply the adaptive filtering algorithm to smooth the multi-source historical performance data to remove noise interference. Then, use the support vector regression model to fit the smoothed data to generate a benchmark performance curve. By comparing the deviation degree between the actual performance curve and the benchmark performance curve, evaluate the potential abnormal risk level of the equipment in combination with a fuzzy logic system.

[0076] For example, after identifying the key factors, engineers applied the adaptive filtering algorithm to smooth the multi-source historical performance data and used the support vector regression model to generate a benchmark performance curve. By comparing the deviation degree between the actual performance curve and the benchmark performance curve, they found that some devices had a relatively high abnormal risk level. These evaluation results provided a basis for subsequent predictions and maintenance suggestions.

[0077] Step 105: According to the unexpected fluctuation patterns, the key factors, and the potential abnormal risk level, use a Bayesian network to predict the floor scrubber production equipment that will fail within a preset time period and generate a maintenance recommendation report.

[0078] In this step, a Bayesian network is a probabilistic graphical model used to represent the dependencies between variables. It can infer the states of unknown variables based on known probability distributions. In this application scenario, the Bayesian network is used to predict the probability of a failure occurring within a specific future time period and generate a maintenance recommendation report.

[0079] In actual operation, first, a Bayesian network model is constructed, and data such as unexpected fluctuation patterns, key factors, and potential abnormal risk levels are input. Then, the model is used to predict the failure probability within a specific future time period and generate a detailed maintenance recommendation report to guide the plant managers to take preventive maintenance measures.

[0080] For example, based on the results of the previous steps, the engineers constructed a Bayesian network model to predict the floor scrubber equipment that may fail within the next month. The report shows that certain equipment has a high failure risk within the next two weeks, and it is recommended to conduct inspections and maintenance immediately. The plant managers took preventive maintenance measures in a timely manner according to this report, avoiding unexpected downtimes.

[0081] Through the comprehensive application of the above five steps, this method realizes the full-process management from real-time monitoring to failure prediction. First, by collecting and analyzing the equipment operation parameters in real time, the periodic and non-periodic components are identified; then, the clustering algorithm and the association rule mining algorithm are used to discover the unexpected fluctuation patterns and their causal relationships, and the key factors are extracted; subsequently, a baseline performance curve is constructed by combining these key factors to evaluate the equipment health status; finally, the future failure probability is predicted through a Bayesian network, and a maintenance recommendation report is generated. This method not only improves the predictability and accuracy of equipment maintenance, but also effectively reduces the unexpected downtime, extends the equipment service life, and greatly improves the production efficiency and economic benefits. Throughout the process, each step is closely connected, forming a complete equipment health management solution.

[0082] To solve the problem that the evaluation of the equipment health status in the prior art is not accurate enough, in some embodiments, in step 104, using the key factors as inputs, a baseline performance curve is constructed by combining the performance data of the same type of components among the floor scrubber production equipment, and based on the deviation degree between the baseline performance curve and the actual performance curve, the floor scrubber production equipment is evaluated by combining a fuzzy logic system to determine the potential abnormal risk level, including:

[0083] Using the key factors as inputs, combining the performance data of the same type of components among the floor scrubber production equipment, applying an adaptive filtering algorithm to smooth the multi-source historical performance data, obtaining the smoothed data, using a support vector regression model to fit the smoothed data, generating a benchmark performance curve, where the multi-source historical performance data includes: the performance data of the same type of components, the actual operation parameters of the current equipment at different time periods, historical fault records, and maintenance logs; by comparing the actual performance curve with the benchmark performance curve, calculating the deviation degree between the actual performance curve and the benchmark performance curve, and using the root mean square error to quantify the deviation degree, obtaining a deviation degree score; according to the deviation degree score, combining a fuzzy logic system to evaluate the floor scrubber production equipment, obtaining an output result, and using a particle swarm optimization algorithm to optimize the output result, obtaining the health status evaluation result of the floor scrubber production equipment; using the health status evaluation result, combining a time series prediction model, predicting the performance change trend of the floor scrubber production equipment within a preset time period, obtaining a prediction result, and analyzing the prediction result to determine the potential abnormal risk level.

[0084] In this embodiment, the adaptive filtering algorithm is a technology that can dynamically adjust its parameters to adapt to the changing environment and is commonly used in the field of signal processing. Support vector regression is an extended form of support vector machines and is used for regression analysis. The multi-source historical performance data includes the performance data of the same type of components, the actual operation parameters of the current equipment at different time periods, historical fault records, and maintenance logs, and these data are used to generate an accurate benchmark performance curve. The root mean square error is a standard measurement method for measuring the difference between the predicted value and the true value and is used to quantify the deviation degree between the actual performance curve and the benchmark performance curve.

[0085] In the embodiment of the present application, first, an adaptive filtering algorithm is applied to smooth the multi-source historical performance data to remove noise interference. Then, a support vector regression model is used to fit the smoothed data to generate a benchmark performance curve. By comparing the deviation degree between the actual performance curve and the benchmark performance curve and using the root mean square error to quantify the deviation degree, a deviation degree score is obtained. According to the deviation degree score, a fuzzy logic system is combined to evaluate the equipment, obtaining an output result, and using a particle swarm optimization algorithm to optimize the output result, and finally obtaining the health status evaluation result of the equipment. Finally, a time series prediction model is combined to predict the performance change trend of the equipment within a specific future time period, and the prediction result is analyzed to determine the potential abnormal risk level.

[0086] The following is a specific embodiment:

[0087] In a large manufacturing plant, in order to improve the accuracy of the health status assessment of floor scrubber production equipment, the above method was adopted. First, they collected the operation parameters, fault records, and maintenance logs of all equipment of the same type in the past year, and used the adaptive filtering algorithm to smooth these data. Then, a support vector regression model was used to generate the baseline performance curve for each piece of equipment. By comparing the operation parameters collected in real time with the baseline performance curve, the deviation degree score was calculated and quantified using the root mean square error. Subsequently, the health status of the equipment was evaluated in combination with a fuzzy logic system, and the particle swarm optimization algorithm was used to further optimize the evaluation results. Based on these evaluation results, the engineers also used a time series prediction model to predict the trend of equipment performance changes in the next month. The results showed that some equipment had a high risk of failure in the next two weeks, and it was recommended to carry out preventive maintenance immediately. This method not only improved the predictability and accuracy of equipment maintenance, but also significantly reduced the unexpected downtime and improved the overall production efficiency. Through this method, the plant management can identify potential problems in advance and formulate effective maintenance strategies, thus ensuring the continuity and stability of the production line.

[0088] In order to further improve the accuracy and robustness of the equipment health status assessment, in some embodiments, the step of evaluating and processing the floor scrubber production equipment according to the deviation degree score in combination with a fuzzy logic system in step 104 to obtain an output result, and using the particle swarm optimization algorithm to optimize the output result to obtain the health status assessment result of the floor scrubber production equipment further includes:

[0089] Use an adaptive neuro - fuzzy inference system to perform fuzzy - logic evaluation processing on the deviation degree score to generate a preliminary evaluation result; based on the preliminary evaluation result, optimize the fuzzy - logic system by combining the genetic algorithm and the differential evolution algorithm to obtain an optimized fuzzy - logic system; input the deviation degree score into the optimized fuzzy - logic system to generate an output result, and use the particle swarm optimization algorithm and the simulated annealing algorithm to optimize the output result to generate an optimized output result; according to the ensemble learning method, comprehensively combine the preliminary evaluation result, the optimized fuzzy - logic system, and the optimized output result to generate a health - status evaluation result of the floor - scrubbing machine production equipment. Optionally, the step of inputting the deviation degree score into the optimized fuzzy - logic system to generate an output result, and using the particle swarm optimization algorithm and the simulated annealing algorithm to optimize the output result to generate an optimized output result includes: using the optimized fuzzy - logic system, taking the deviation degree score as input data, performing fuzzy - inference processing on the deviation degree score, and outputting a preliminary equipment health - status evaluation result; using the particle swarm optimization algorithm to perform a global search on the preliminary equipment health - status evaluation result to identify the optimal solution, and based on the optimal solution, optimizing the preliminary equipment health - status evaluation result to obtain the result of the particle swarm optimization algorithm; according to the result of the particle swarm optimization algorithm, combining with the simulated annealing algorithm, further optimizing the preliminary equipment health - status evaluation result to generate an optimized output result.

[0090] In this embodiment, the adaptive neuro - fuzzy inference system is an intelligent system that combines neural networks and fuzzy logic and is used to process complex data patterns. The genetic algorithm and the differential evolution algorithm are two commonly used global optimization algorithms that search for the optimal solution by simulating the biological evolution process. The particle swarm optimization algorithm is an optimization method based on swarm intelligence that simulates the foraging behavior of birds to search for the global optimal solution. The simulated annealing algorithm is a probabilistic technique used to find an approximate optimal solution in a large - scale search space. These methods work together to improve the accuracy and reliability of the equipment health - status evaluation.

[0091] In the embodiments of the present application, first, the adaptive neuro-fuzzy inference system is used to perform fuzzy logic evaluation processing on the deviation degree score to generate a preliminary evaluation result. Then, based on the preliminary evaluation result, in a manner combining the genetic algorithm and the differential evolution algorithm, the fuzzy logic system is optimized to obtain an optimized fuzzy logic system. The deviation degree score is input into the optimized fuzzy logic system to generate an output result, and the particle swarm optimization algorithm and the simulated annealing algorithm are used to optimize this output result to generate an optimized output result. Finally, according to the ensemble learning method, the preliminary evaluation result, the optimized fuzzy logic system, and the optimized output result are comprehensively used to generate the health state evaluation result of the floor scrubber production equipment. Specifically, the optimized fuzzy logic system is used to perform fuzzy inference processing on the deviation degree score to output a preliminary equipment health state evaluation result; the particle swarm optimization algorithm is used to perform a global search on the preliminary result to identify the optimal solution and optimize the preliminary result; according to the result of the particle swarm optimization algorithm, the preliminary result is optimized again in combination with the simulated annealing algorithm to generate the final optimized output result.

[0092] The following is a specific embodiment:

[0093] In a large manufacturing plant, in order to improve the accuracy of the health state evaluation of the floor scrubber production equipment, the above method is adopted. First, they used the adaptive neuro-fuzzy inference system to perform fuzzy logic evaluation processing on the deviation degree score of each device to generate a preliminary evaluation result. Then, based on the preliminary evaluation result, the engineers optimized the fuzzy logic system by combining the genetic algorithm and the differential evolution algorithm to obtain a more accurate optimized fuzzy logic system. Subsequently, the deviation degree score was input into the optimized fuzzy logic system to generate a preliminary equipment health state evaluation result. In order to further improve the accuracy of the evaluation result, the engineers used the particle swarm optimization algorithm to perform a global search on the preliminary result, found the optimal solution, and optimized the preliminary result based on this optimal solution. Finally, in combination with the simulated annealing algorithm, the preliminary result was optimized again to generate the final optimized output result. The results showed that some devices had a high risk of failure in the next two weeks, and it was recommended to perform preventive maintenance immediately. This method not only significantly improved the accuracy and robustness of the equipment health state evaluation, but also effectively reduced the unexpected downtime and improved the overall production efficiency. Through this method, the factory management can identify potential problems in advance, formulate effective maintenance strategies, and thus ensure the continuity and stability of the production line.

[0094] In order to further improve the accuracy of identifying the causal relationship and co-occurrence pattern in the unexpected fluctuation pattern, in some embodiments, in step 103, the association rule mining algorithm is used to perform causal relationship and co-occurrence pattern discovery processing on the unexpected fluctuation pattern to obtain key factors, including:

[0095] Construct a data set using the unexpected fluctuation pattern, perform dimensionality reduction on the data set using principal component analysis to obtain an optimized data set, use the association rule mining algorithm to scan the optimized data set to identify frequent item sets; use Granger causality test to detect the frequent item sets, obtain the causal relationship of the operating parameters, and extract the co-occurrence pattern from the frequent item sets. Combine the causal relationship and the co-occurrence pattern to obtain a key set; use lift, Jaccard similarity coefficient, and Sørensen-Dice coefficient as evaluation indicators to evaluate the causal relationship and co-occurrence pattern of the key set, and generate a comprehensive scoring result; according to the comprehensive scoring result, screen out candidate key factors from the frequent item sets, and use the random forest model to perform importance scoring on the candidate key factors to obtain key factors. Optionally, using lift, Jaccard similarity coefficient, and Sørensen-Dice coefficient as evaluation indicators to evaluate the causal relationship and co-occurrence pattern of the key set and generate a comprehensive scoring result includes: calculating the lift of the causal relationship and co-occurrence pattern in the key set to obtain a correlation strength score; using the Jaccard similarity coefficient to perform similarity evaluation on the causal relationship and co-occurrence pattern in the key set, calculating the similarity degree of the causal relationship and co-occurrence pattern in the key set to obtain a similarity score; using the Sørensen-Dice coefficient to perform uniqueness and importance evaluation on the causal relationship and the co-occurrence pattern, calculating the uniqueness score of the causal relationship and the co-occurrence pattern to obtain a uniqueness score; combining the correlation strength score, the similarity score, and the uniqueness score to perform comprehensive scoring on each causal relationship and co-occurrence pattern in the key set to obtain a comprehensive scoring result.

[0096] In this embodiment, principal component analysis is a technique for dimensionality reduction that reduces the dimension of a data set by projecting the data onto a new coordinate system while trying to retain the main features of the original data as much as possible. The association rule mining algorithm is used to discover frequent item sets, that is, combinations of items that co-occur in a data set. Granger causality test is a statistical method used to detect causal relationships in time series data. Lift, Jaccard similarity coefficient, and Sørensen-Dice coefficient are evaluation indicators used to evaluate the correlation strength, similarity, and uniqueness of causal relationships and co-occurrence patterns, respectively. The random forest model is an ensemble learning method that makes predictions by constructing multiple decision trees and determines the importance of variables based on the importance scores of each tree.

[0097] In the embodiments of the present application, first, a data set is constructed using the unexpected fluctuation pattern, and principal component analysis is used to reduce its dimension, obtaining an optimized data set. Then, the association rule mining algorithm is applied to scan the optimized data set to identify frequent item sets. Next, the Granger causality test is used to detect these frequent item sets, and the causal relationships and co-occurrence patterns of the operating parameters are extracted to form a key set. The lift, Jaccard similarity coefficient, and Sørensen-Dice coefficient are used as evaluation indicators to perform a comprehensive scoring process on the key set, generating a comprehensive scoring result. Finally, according to the comprehensive scoring result, candidate key factors are screened out from the frequent item sets, and the random forest model is used to perform an importance scoring process on these candidate key factors, ultimately obtaining the key factors.

[0098] The following is a specific embodiment:

[0099] In a large manufacturing plant, in order to more accurately identify the unexpected fluctuation patterns and their underlying causal relationships during the operation of floor scrubber production equipment, the above method is adopted. First, they constructed a data set containing operating parameters such as temperature, pressure, and vibration using the unexpected fluctuation pattern, and reduced the dimension of this data set through principal component analysis, obtaining an optimized data set. Then, the Apriori algorithm was applied to scan the optimized data set to identify frequent item sets. Next, the Granger causality test was used to detect these frequent item sets, and the causal relationships and co-occurrence patterns between the operating parameters were extracted to form a key set. To evaluate the quality of these key sets, the engineers calculated the lift, Jaccard similarity coefficient, and Sørensen-Dice coefficient, obtaining a comprehensive scoring result. Based on these scoring results, they screened out several candidate key factors from the frequent item sets, and used the random forest model to perform an importance scoring process on these candidate factors, ultimately determining the key factors affecting the health status of the equipment. The results showed that certain specific combinations of operating parameters were significantly associated with the occurrence of equipment failures, providing an important reference basis for subsequent preventive maintenance. This method not only improves the accuracy of identifying causal relationships and co-occurrence patterns, but also effectively enhances the predictability and efficiency of equipment maintenance.

[0100] In order to further improve the accuracy of identifying unexpected fluctuation patterns, in some embodiments, the step of identifying the unexpected fluctuation patterns that appear in the operating parameters using the clustering algorithm according to the periodic and non-periodic components in step 102 includes:

[0101] Based on the periodic and aperiodic components, a clustering algorithm is used to classify the operating parameters, identify multiple operating modes and abnormal modes, and obtain a clustering result; according to the operating modes and the abnormal modes, the clustering result is obtained; by comparing the clustering result with the known normal operating mode, the operating parameters that do not conform to the normal operating mode are identified and marked as unexpected fluctuation modes.

[0102] In this embodiment, the clustering algorithm is an unsupervised learning method used to divide the objects in a dataset into multiple groups or clusters such that the objects within the same group are more similar to each other than those in other groups. The periodic and aperiodic components refer to the regular and irregular change parts in the device operating parameters respectively. The normal operating mode refers to the typical operating parameter characteristics exhibited by the device in the normal working state. By comparing the clustering result with the known normal operating mode, the operating parameters that do not conform to the normal mode can be identified and marked as unexpected fluctuation modes.

[0103] In the embodiment of the present application, first, based on the periodic and aperiodic components, a clustering algorithm is used to classify the operating parameters, identify multiple operating modes and abnormal modes, and obtain a clustering result. Then, by comparing these clustering results with the known normal operating mode, the operating parameters that do not conform to the normal operating mode are identified and marked as unexpected fluctuation modes. This method can effectively screen out potential problem points from a large amount of data, providing a basis for subsequent fault diagnosis and maintenance.

[0104] The following is a specific example:

[0105] In a large manufacturing plant, in order to more accurately identify the unexpected fluctuation modes in the operation of floor scrubbing machine production equipment, the above method is adopted. First, they collected the operating parameters such as temperature, pressure, and vibration of each device in real time through a sensor network, and used signal processing technology to decompose these parameters into periodic and aperiodic components. Then, the K-means clustering algorithm was applied to classify these components, identify multiple operating modes and abnormal modes, and obtain a clustering result. Subsequently, the engineers compared these clustering results with the known normal operating mode, and identified and marked the operating parameters that did not conform to the normal mode as unexpected fluctuation modes. For example, some devices showed vibration values significantly higher than the normal level during a specific period, which might be caused by mechanical component wear or other potential problems. Through this method, the engineers can not only timely detect the abnormal conditions in the device operation, but also take preventive maintenance measures in advance to avoid production line interruption caused by sudden failures. This strategy significantly improves the predictability and efficiency of device maintenance, ensuring the continuity and stability of production.

[0106] This application takes into account that in the field of industrial equipment monitoring and maintenance, accurately evaluating the health status of equipment and predicting potential failures is crucial for improving production efficiency, reducing downtime, and extending the service life of equipment. Traditional methods usually rely on simple threshold alarm mechanisms or static benchmark performance curves, which are difficult to adapt to complex and changing actual working conditions, easily leading to false alarms or missed alarms and affecting the accuracy of maintenance decisions. In addition, existing solutions are less efficient when dealing with large-scale data sets and cannot respond in real time to rapidly changing equipment states. Therefore, a new alternative solution is proposed, which includes:

[0107] By comparing the actual performance curve with the benchmark performance curve, calculating the degree of deviation between the actual performance curve and the benchmark performance curve, and quantifying the degree of deviation using the root mean square error to obtain a deviation degree score, including:

[0108] By comparing the actual value in the actual performance curve with the benchmark value in the benchmark performance curve, calculating the degree of deviation between the actual performance curve and the benchmark performance curve, where the weighted non-linear difference between the actual value and the benchmark value is expressed as:

[0109] ;

[0110] Where represents the weighted non-linear difference between the actual value and the benchmark value at the -th time point, is the attenuation coefficient, is the current time, is the time point 's time stamp, represents the actual operating parameter value at the -th time point, represents the ideal reference value at the -th time point;

[0111] The following is a detailed explanation of each parameter:

[0112] represents the weighted non-linear difference between the actual value and the benchmark value at the -th time point. This value is used to measure the degree of deviation between the actual value of the equipment operating parameter and the ideal reference value at a specific time point, and takes into account the time decay effect.

[0113] represents the dynamic weight function, indicating the importance of the data at the -th time point. As the time distance from the current time increases, this weight will gradually decrease, reflecting that earlier data has less impact on the current evaluation. The acquisition method is by setting the attenuation coefficient and the current time Calculate the timestamps for each time point of the weights.

[0114] represents the decay coefficient, which determines the rate at which the time weight changes over time. A larger value will cause the weight to decline faster, meaning more emphasis is placed on recent data. The value is obtained by determining a suitable value based on the requirements of the specific application scenario and historical data analysis.

[0115] represents the current time, which is used to calculate the time difference between each time point and the current moment. The value is obtained by directly reading from the system clock or real-time monitoring system.

[0116] represents the timestamp of the

[0117] th time point, which is used to calculate the time difference relative to the current time. The value is obtained by extracting the timestamps of each time point from historical data records.

[0118] represents the th actual operating parameter value of the time point, which represents the actual operating state of the device at that time point. The value is obtained by real-time acquisition from the sensor network or other data acquisition systems.

[0119] represents the th ideal reference value (i.e., the value on the benchmark performance curve) of the time point, which represents the operating parameter value that the device should reach under normal operating conditions. The value is obtained based on the benchmark performance curve established from historical data, experimental tests, or industry standards.

[0120] represents the sum of the dynamic weights of all time points, which is used for normalization to ensure that the total weight of data at different time points is 1. The value is obtained by traversing and calculating the timestamps of all time points and accumulating the dynamic weights of each time point. and accumulating the dynamic weights of each time point.

[0121] The reasons for the design of each item are introduced as follows:

[0122] Dynamic weight function The reason for the design of introducing a dynamic weight function is to assign higher weights to the most recent data, reflecting its importance for the current evaluation. This approach can reduce the impact of old data on the evaluation results and improve the response speed and accuracy of the model.

[0123] Nonlinear transformation function The reason for the design of using a nonlinear transformation function is that it can enhance the sensitivity to outliers, making even small deviations amplified, thus making it easier to identify potential problems.

[0124] Denominator The reason for the design is that through normalization, it is ensured that the sum of the data weights at different time points is 1, avoiding distortion of the results caused by overly large or small weights.

[0125] In the formula, multiplying each term is for the dynamic weight and the result of the nonlinear transformation are multiplied. The purpose is to combine the time decay effect and the difference between the actual value and the reference value, and comprehensively evaluate the importance of each time point and its degree of deviation.

[0126] Based on the result of the weighted nonlinear difference, calculate the generalized mean of the weighted squared differences at all time points. The calculation formula is as follows:

[0127] ;

[0128] Where, represents the generalized form of the weighted mean square error. By introducing a dynamic exponent the sensitivity to errors is adjusted, represents the sum of the dynamic weights at all time points, is a tuning parameter used to control the amplitude of the dynamic exponent, is the angular frequency;

[0129] The following is a detailed explanation of each parameter:

[0130] represents the generalized form of the weighted mean square error, which is used to measure the deviation between the actual performance curve and the reference performance curve. The acquisition method is calculated through the above formula.

[0131] represents the absolute value of the weighted nonlinear difference between the actual value and the reference value at the

[0132] represents the dynamic exponent, which is used to adjust the sensitivity to errors. The acquisition method is based on the tuning parameter and the angular frequency to calculate the current time The dynamic index.

[0133] Represents the adjustment parameter used to control the amplitude of the dynamic index. The acquisition method is set according to specific application requirements and is usually determined through experiments or experience.

[0134] Represents the angular frequency, which determines the periodic change speed of the dynamic index. The acquisition method is determined based on the operating characteristics of the device and historical data analysis.

[0135] Represents the sum of the dynamic weights at all time points and is used for normalization. The acquisition method is to traverse and calculate the timestamps at all time points and accumulate the dynamic weights at each time point.

[0136] The following is an introduction to the design reasons for each sub-item:

[0137] Dynamic index The design reason for the dynamic index is that it is designed to adapt to different requirements for error sensitivity in different time periods. By introducing periodic changes ( , the sensitivity to errors can be adjusted in different time periods, making the evaluation more flexible and accurate.

[0138] Denominator The design reason for the denominator is to ensure that the sum of the data weights at different time points is 1 through normalization processing, avoiding result distortion caused by excessive or too small weights.

[0139] In the formula, by multiplying each sub-item, the dynamic weight is multiplied by the absolute value of the weighted non-linear difference to the power of the dynamic index. The purpose is to combine the time decay effect, the magnitude of the error, and its sensitivity to comprehensively evaluate the importance of each time point and its deviation degree.

[0140] Based on the result of the generalized mean, calculate the weighted root mean square error to obtain the deviation degree score. The calculation formula is as follows:

[0141] ;

[0142] Among them, represents the weighted root mean square error, is the adjustment parameter used to control the amplitude of the dynamic adjustment factor.

[0143] The following is a detailed explanation of each parameter:

[0144] Denotes the weighted root mean square error, which is used to quantify the deviation between the actual performance curve and the benchmark performance curve. It is obtained by calculating according to the above formula.

[0145] Denotes the adjustment parameter, which is used to control the amplitude of the dynamic adjustment factor. A larger value will make the influence of the dynamic adjustment factor greater. It is obtained by determining the appropriate value according to the requirements of the specific application scenario and the analysis of historical data.

[0146] The reasons for each sub - item design are introduced as follows:

[0147] Dynamic adjustment factor The reason for its design is that introducing the dynamic adjustment factor is to adapt to the changing trend of the equipment health state in different time periods. By using the sine function, the sensitivity of the evaluation result can be adjusted in different time periods, making the evaluation more flexible and accurate.

[0148] Weighted mean square error The reason for its design is is the core index to measure the deviation between the actual performance curve and the benchmark performance curve. By introducing dynamic weights and dynamic exponents, the evaluation result can adapt to complex and changeable actual working conditions.

[0149] Dynamic weight function The reason for its design is that introducing the dynamic weight function is to give higher weights to the most recent data, reflecting its importance to the current evaluation. This method can reduce the influence of old data on the evaluation result and improve the response speed and accuracy of the model.

[0150] Denominator The reason for its design is that by accumulating and normalizing the dynamic weights at all time points, it ensures that the sum of the data weights at different time points is 1, avoiding the result distortion caused by too large or too small weights, thus improving the stability and reliability of the evaluation result.

[0151] In the formula, by multiplying each sub - item, the dynamic adjustment factor is multiplied by the square root of the weighted mean square error, aiming to combine the periodic change and the magnitude and sensitivity of the error to comprehensively evaluate the importance and deviation degree of each time point.

[0152] The following is a specific example:

[0153] In a large manufacturing plant, engineers used the above - mentioned method to evaluate the health state of the floor - scrubbing machine production equipment. Suppose we have the following values:

[0154] The current time is Hour, time point Hour, Hour, Hour, attenuation coefficient , angular frequency , adjustment parameter , actual operating parameter value: , reference value: .

[0155] First, calculate the weighted non - linear difference at each time point :

[0156] ;

[0157] Then, calculate the generalized mean :

[0158] ;

[0159] Finally, calculate the weighted root - mean - square error :

[0160] ;

[0161] It can be seen from the calculation results that the weighted root - mean - square error is approximately 1.58. This means that there is a significant deviation between the actual performance curve and the reference performance curve, indicating that there may be some problems or abnormalities in the equipment. Based on this result, engineers can take corresponding preventive maintenance measures, such as checking the operating condition of the equipment within a specific time period and troubleshooting the possible causes of abnormalities (such as mechanical component wear, sensor failure, etc.), so as to avoid potential failures and ensure the continuity and stability of production. This method not only improves the accuracy of equipment health status assessment but also enhances the predictability and effectiveness of maintenance strategies.

[0162] Figure 2 FIG. is a schematic structural diagram of an abnormal prediction system for a floor - mopping machine production equipment based on AI provided by an embodiment of the present application. As Figure 2 shown, the system includes:

[0163] A receiving module 21, configured to receive the operating parameters of the floor - mopping machine production equipment from real - time monitoring, compare the operating parameters with the historical data of the operation of the floor - mopping machine production equipment, and obtain the periodic and non - periodic components in the operating parameters;

[0164] An identification module 22, configured to identify an unexpected fluctuation pattern occurring in the operation parameters according to the periodic and aperiodic components by using a clustering algorithm;

[0165] A utilization module 23, configured to perform causality and co-occurrence pattern discovery processing on the unexpected fluctuation pattern by using an association rule mining algorithm to obtain key factors;

[0166] A construction module 24, configured to use the key factors as inputs, combine the performance data of the same type of components among the floor scrubber production devices, construct a benchmark performance curve, and evaluate the floor scrubber production devices in combination with a fuzzy logic system based on the deviation degree between the benchmark performance curve and the actual performance curve to determine the potential abnormal risk level;

[0167] A prediction module 25, configured to predict the floor scrubber production devices that will fail within a preset time period by using a Bayesian network according to the unexpected fluctuation pattern, the key factors, and the potential abnormal risk level, and generate a maintenance recommendation report.

[0168] Figure 2 The above-mentioned AI-based floor scrubber production device anomaly prediction system can execute Figure 1 The AI-based floor scrubber production device anomaly prediction method described in the illustrated embodiment, and its implementation principle and technical effects will not be elaborated. For the above-mentioned AI-based floor scrubber production device anomaly prediction system in the embodiment, the specific manners in which each module and unit perform operations have been described in detail in the embodiment related to the method, and will not be elaborated here.

[0169] In a possible design, Figure 2 The AI-based floor scrubber production device anomaly prediction system in the illustrated embodiment can be implemented as a computing device, such as Figure 3 shown, and the computing device may include a storage component 31 and a processing component 32;

[0170] The storage component 31 stores one or more computer instructions, and among them, the one or more computer instructions are called and executed by the processing component 32.

[0171] The processing component 32 performs the above Figure 1 AI-based floor scrubber production device anomaly prediction method in the embodiment.

[0172] Among them, the processing component 32 may include one or more processors to execute computer instructions to complete all or part of the steps in the above methods. Of course, the processing component may also be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors or other electronic components for executing the above methods.

[0173] The storage component 31 is configured to store various types of data to support the operation of the terminal. The storage component may be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0174] Of course, the computing device may also necessarily include other components, such as input / output interfaces, display components, communication components, etc.

[0175] The input / output interface provides an interface between the processing component and the peripheral interface module, and the above peripheral interface module may be an output device, an input device, etc.

[0176] The communication component is configured to facilitate communication between the computing device and other devices in a wired or wireless manner, etc.

[0177] Among them, the computing device may be a physical device or an elastic computing host provided by a cloud computing platform, etc. At this time, the computing device may refer to a cloud server, and the above processing component, storage component, etc. may be basic server resources leased or purchased from a cloud computing platform.

[0178] The embodiment of the present application also provides a computer storage medium storing a computer program, and when the computer program is executed by a computer, it can implement the above Figure 1 AI-based abnormal prediction method for floor scrubber production equipment shown in the embodiment.

[0179] Those skilled in the art can clearly understand that for the convenience and conciseness of description, the specific working processes of the above-described systems, devices, and units can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0180] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. A person of ordinary skill in the art can understand and implement it without creative work.

[0181] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on such an understanding, the essence of the above technical solution, or the part that contributes to the prior art, can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0182] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, and are not intended to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. The AI-based method for predicting abnormalities in floor scrubbing machine production equipment is characterized by: include: Receiving operating parameters of a floor scrubber production device from real-time monitoring, comparing the operating parameters with historical data from the operation of the floor scrubber production device, and obtaining periodic and non-periodic components in the operating parameters; identifying unexpected fluctuation patterns occurring in the operating parameters using a clustering algorithm based on the periodic and non-periodic components; Using association rule mining algorithms, the causal relationship and co-occurrence pattern discovery processing of the unexpected fluctuation pattern is performed to obtain key factors; Using the key factors as input, combined with the performance data of the same type of components among floor scrubber production equipment, a baseline performance curve is constructed, and based on the degree of deviation between the baseline performance curve and the actual performance curve, the floor scrubber production equipment is evaluated in combination with a fuzzy logic system to determine the potential abnormal risk level; Based on the unexpected fluctuation pattern, the key factors and the potential abnormal risk level, a Bayesian network is used to predict the floor scrubbing machine production equipment that will fail within a preset time period, and a maintenance recommendation report is generated; The method uses the key factors as inputs, combines the performance data of the same type of components among floor scrubber production equipment to build a benchmark performance curve, evaluates the floor scrubber production equipment based on the degree of deviation between the benchmark performance curve and the actual performance curve, and combines the fuzzy logic system to determine the potential abnormal risk level, including: Using the key factors as input, combined with the performance data of the same type of components among floor scrubbing machine production equipment, an adaptive filtering algorithm is applied to smooth the historical performance data from multiple sources to obtain smoothed data, and a support vector regression model is used to fit the smoothed data to generate a benchmark performance curve, wherein the multi-source historical performance data includes: performance data of the same type of components, actual operating parameters of the current equipment in different time periods, historical fault records, and maintenance logs; By comparing the actual performance curve with the benchmark performance curve, calculating the degree of deviation between the actual performance curve and the benchmark performance curve, and quantifying the degree of deviation using a root mean square error to obtain a degree of deviation score; According to the deviation degree score, the floor scrubber production equipment is evaluated and processed in combination with a fuzzy logic system to obtain an output result, and the output result is optimized using a particle swarm optimization algorithm to obtain a health status evaluation result of the floor scrubber production equipment; Using the health status assessment results, combined with the time series prediction model, the performance change trend of the floor scrubbing machine production equipment within a preset time period is predicted and processed to obtain a prediction result, and the prediction result is analyzed to determine the potential abnormal risk level; According to the deviation degree score, the floor scrubber production equipment is evaluated and processed in combination with the fuzzy logic system to obtain an output result, and the output result is optimized by the particle swarm optimization algorithm to obtain the health status evaluation result of the floor scrubber production equipment, including: Using an adaptive neuro-fuzzy inference system to perform fuzzy logic evaluation on the deviation degree score to generate a preliminary evaluation result; Based on the preliminary evaluation results, the fuzzy logic system is optimized by combining the genetic algorithm with the differential evolution algorithm to obtain an optimized fuzzy logic system; Inputting the deviation degree score into the optimized fuzzy logic system to generate an output result, and optimizing the output result by using a particle swarm optimization algorithm and a simulated annealing algorithm to generate an optimized output result; According to the integrated learning method, the preliminary evaluation results, the optimized fuzzy logic system and the optimized output results are integrated to generate the health status evaluation results of the floor scrubbing machine production equipment.

2. The method according to claim 1, characterized in that The step of inputting the deviation degree score into the optimized fuzzy logic system to generate an output result, and optimizing the output result by using a particle swarm optimization algorithm and a simulated annealing algorithm to generate an optimized output result includes: Utilizing the optimized fuzzy logic system, taking the deviation degree score as input data, performing fuzzy reasoning processing on the deviation degree score, and outputting a preliminary equipment health status assessment result; Using a particle swarm optimization algorithm to perform a global search on the preliminary equipment health status assessment results, identify an optimal solution, and based on the optimal solution, optimize the preliminary equipment health status assessment results to obtain a result of the particle swarm optimization algorithm; According to the result of the particle swarm optimization algorithm, combined with the simulated annealing algorithm, the preliminary equipment health status assessment result is optimized again to generate an optimized output result.

3. The method according to claim 1, characterized in that The association rule mining algorithm is used to perform causal relationship and co-occurrence pattern discovery processing on the unexpected fluctuation pattern to obtain key factors, including: Constructing a data set using the unexpected fluctuation pattern, performing dimensionality reduction processing on the data set using principal component analysis to obtain an optimized data set, and scanning the optimized data set using an association rule mining algorithm to identify frequent item sets; Using Granger causality test to detect the frequent item sets, obtain the causal relationship of the operation parameters, extract the co-occurrence pattern from the frequent item sets, and combine the causal relationship and the co-occurrence pattern to obtain a key set; Using lift, Jaccard similarity coefficient and Sorenson-Dees coefficient as evaluation indicators, the causal relationship and co-occurrence pattern of the key set are evaluated to generate a comprehensive scoring result; According to the comprehensive scoring result, candidate key factors are screened out from the frequent item set, and the candidate key factors are scored according to their importance using a random forest model to obtain key factors.

4. The method according to claim 3, characterized in that The lift, Jaccard similarity coefficient and Sorenson-Dees coefficient are used as evaluation indicators to evaluate the causal relationship and co-occurrence pattern of the key set and generate a comprehensive scoring result, including: Calculating the lift of causal relationships and co-occurrence patterns in the key set to obtain a correlation strength score; Using the Jaccard similarity coefficient, similarity evaluation is performed on the causal relationship and the co-occurrence pattern in the key set, and the similarity degree of the causal relationship and the co-occurrence pattern in the key set is calculated to obtain a similarity score; Using the Sorenson-Dees coefficient, the causal relationship and the co-occurrence pattern are evaluated for uniqueness and importance, and the uniqueness scores of the causal relationship and the co-occurrence pattern are calculated to obtain a uniqueness score; Combined with the correlation strength score, the similarity score and the uniqueness score, a comprehensive scoring process is performed on each causal relationship and co-occurrence pattern in the key set to obtain a comprehensive scoring result.

5. The method according to claim 1, characterized in that The step of identifying unexpected fluctuation patterns in the operating parameters using a clustering algorithm based on the periodic and non-periodic components includes: Based on the periodicity and the non-periodicity components, the operating parameters are classified using a clustering algorithm to identify a plurality of operating modes and abnormal modes, and a clustering result is obtained according to the operating modes and the abnormal modes; By comparing the clustering results with known normal operating modes, operating parameters that do not conform to the normal operating modes are identified and marked as unexpected fluctuation modes.

6. An AI-based abnormality prediction system for floor scrubbing machine production equipment, characterized in that: include: A receiving module, used for receiving operating parameters of a floor scrubber production device from real-time monitoring, comparing the operating parameters with historical data from the operation of the floor scrubber production device, and obtaining periodic and non-periodic components in the operating parameters; an identification module, configured to identify unexpected fluctuation patterns occurring in the operating parameters using a clustering algorithm based on the periodic and non-periodic components; A module is used to utilize an association rule mining algorithm to perform causal relationship and co-occurrence pattern discovery processing on the unexpected fluctuation pattern to obtain key factors; A construction module is used to use the key factors as inputs, combine the performance data of the same type of components among floor scrubbing machine production equipment, construct a benchmark performance curve, and evaluate the floor scrubbing machine production equipment in combination with a fuzzy logic system based on the degree of deviation between the benchmark performance curve and the actual performance curve to determine the potential abnormal risk level; A prediction module, for predicting the floor scrubbing machine production equipment that will fail within a preset time period using a Bayesian network according to the unexpected fluctuation pattern, the key factors and the potential abnormal risk level, and generating a maintenance recommendation report; The method uses the key factors as inputs, combines the performance data of the same type of components among floor scrubber production equipment to build a benchmark performance curve, evaluates the floor scrubber production equipment based on the degree of deviation between the benchmark performance curve and the actual performance curve, and combines the fuzzy logic system to determine the potential abnormal risk level, including: Using the key factors as input, combined with the performance data of the same type of components among floor scrubbing machine production equipment, an adaptive filtering algorithm is applied to smooth the historical performance data from multiple sources to obtain smoothed data, and a support vector regression model is used to fit the smoothed data to generate a benchmark performance curve, wherein the multi-source historical performance data includes: performance data of the same type of components, actual operating parameters of the current equipment in different time periods, historical fault records, and maintenance logs; By comparing the actual performance curve with the benchmark performance curve, calculating the degree of deviation between the actual performance curve and the benchmark performance curve, and quantifying the degree of deviation using a root mean square error to obtain a degree of deviation score; According to the deviation degree score, the floor scrubber production equipment is evaluated and processed in combination with a fuzzy logic system to obtain an output result, and the output result is optimized using a particle swarm optimization algorithm to obtain a health status evaluation result of the floor scrubber production equipment; Using the health status assessment results, combined with the time series prediction model, the performance change trend of the floor scrubbing machine production equipment within a preset time period is predicted and processed to obtain a prediction result, and the prediction result is analyzed to determine the potential abnormal risk level; According to the deviation degree score, the floor scrubber production equipment is evaluated and processed in combination with the fuzzy logic system to obtain an output result, and the output result is optimized by the particle swarm optimization algorithm to obtain the health status evaluation result of the floor scrubber production equipment, including: Using an adaptive neuro-fuzzy inference system to perform fuzzy logic evaluation on the deviation degree score to generate a preliminary evaluation result; Based on the preliminary evaluation results, the fuzzy logic system is optimized by combining the genetic algorithm with the differential evolution algorithm to obtain an optimized fuzzy logic system; Inputting the deviation degree score into the optimized fuzzy logic system to generate an output result, and optimizing the output result by using a particle swarm optimization algorithm and a simulated annealing algorithm to generate an optimized output result; According to the integrated learning method, the preliminary evaluation results, the optimized fuzzy logic system and the optimized output results are integrated to generate the health status evaluation results of the floor scrubbing machine production equipment.

7. A computing device, characterized in that It includes a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement the AI-based floor cleaning machine production equipment abnormality prediction method as described in any one of claims 1 to 5.

8. A computer storage medium, characterized in that: A computer program is stored, and when the computer program is executed by a computer, the abnormality prediction method for floor scrubbing machine production equipment based on AI as described in any one of claims 1 to 5 is implemented.

Images