Intelligent evaluation method and system for oxidation resistance of knitting oiling agent and storage medium

Abstract

The invention relates to the technical field of production control, and discloses an intelligent evaluation method and system for the oxidation resistance of knitting oil and a storage medium. The method comprises the following steps: respectively acquiring historical operation parameters of knitting mechanical equipment and historical oxidation products of knitting oil within preset time by using a first sensor and a second sensor, and defining the historical operation parameters and the historical oxidation products as first sequence data and second sequence data; dividing the oxidation process of the knitting oil agent into a plurality of target stages based on the first sequence data and the second sequence data, and establishing a prediction model for each target stage to predict the oxidation resistance of the knitting oil agent on the knitting mechanical equipment; operation parameters of the knitting mechanical equipment in the real-time production process are obtained, the current oxidation stage of the knitting oil agent is positioned, and an anti-oxidation index is output based on the corresponding prediction model; if the current oxidation stage is the middle stage or the later stage, a self-adaptive maintenance period is generated to maintain the knitting oil, and the intelligent evaluation efficiency of the oxidation resistance of the knitting oil is improved.

Description

Technical Field

[0001] The present application relates to the field of production control technology, and in particular to a method, system and storage medium for intelligently evaluating the antioxidant properties of knitting oils. Background Art

[0002] With the rapid development of the textile industry, knitting oils play a vital role in the knitting production process. Knitting oils are primarily used for lubrication, antistatic properties, and fiber friction reduction, thereby improving the operating efficiency of knitting machinery and product quality. However, the performance of knitting oils is affected by various factors during use, with antioxidant properties being a key factor affecting their service life and performance stability.

[0003] Traditional methods for evaluating the antioxidant properties of knitting oils rely primarily on chemical analysis and laboratory testing. However, the antioxidant properties of knitting oils are influenced not only by their chemical composition but also by the operating environment (such as temperature, humidity, and mechanical shear). Traditional evaluation methods often overlook the combined influence of these external factors, resulting in significant deviations between the evaluation results and actual usage.

[0004] For example, Chinese patent application CN104342253A relates to an antioxidant knitting machine oil composition and its uses, primarily addressing the existing problem of knitting machine oil compositions lacking both anti-wear and antioxidant properties. This invention utilizes the following components, by weight: a) 0.2-1 parts of a phosphite extreme pressure anti-wear agent; b) 0.1-1 parts of an antioxidant; c) 0.05-0.2 parts of a rust inhibitor; and d) 98-99.5 parts of a base oil. This technical solution effectively addresses this issue and can be used for lubricating seamless knitting equipment. However, there is no intelligent method for evaluating the antioxidant properties of knitting oils during production.

[0005] In recent years, the rapid development of sensor technology, data analysis technology, and artificial intelligence technology has provided new ideas and methods for evaluating the antioxidant properties of knitting oils. By introducing intelligent evaluation methods, rapid, accurate, and real-time assessment of the antioxidant properties of knitting oils can be achieved. These methods can also be effectively integrated with production management systems, providing strong support for optimizing knitting production processes and controlling quality. Therefore, there is a need for an intelligent evaluation method for the antioxidant properties of knitting oils that can be combined with actual operational data from the production process to evaluate the antioxidant properties of knitting oils and improve production efficiency. Summary of the Invention

[0006] In order to solve the above technical problems, the present application provides an intelligent evaluation method, system and storage medium for the antioxidant properties of knitting oils, which are used to improve the evaluation efficiency of the antioxidant properties of knitting oils.

[0007] In a first aspect, the present application provides a method for intelligently evaluating the antioxidant properties of a knitting oil, the method comprising:

[0008] Step S1: installing a plurality of first sensors and second sensors, wherein the first sensors collect historical operating parameters of the knitting machine within a preset time, which is defined as first sequence data, and the second sensors collect historical oxidation products of the knitting oil within the preset time, which is defined as second sequence data;

[0009] Step S2: dividing the oxidation process of the knitting oil into multiple target stages based on the first sequence data and the second sequence data, establishing a corresponding prediction model for each target stage, the prediction model being used to predict the antioxidant performance of the knitting oil on the knitting machine equipment within the target stage and outputting an antioxidant index;

[0010] Step S3: obtaining current operating parameters of the knitting machine equipment in a real-time production process, locating the current oxidation stage of the knitting oil based on the current operating parameters, and outputting a corresponding antioxidant index based on a corresponding prediction model;

[0011] Step S4: If the current oxidation stage is the middle stage or the late stage, an adaptive maintenance cycle is generated, and the use of the knitting oil is optimized within the adaptive maintenance cycle.

[0012] In combination with the first aspect, in a first implementation of the first aspect of the present application, the oxidation process of the knitting oil is divided into multiple target stages, including:

[0013] First feature data and second feature data are extracted from the first sequence data and the second sequence data, respectively. The first feature data includes time series features, frequency domain features, peak features, and extreme value features of historical operating parameters. The second feature data includes concentration change features, accumulation features, and fluctuation features of historical oxidation products. The correlation between the first feature data and the second feature data and the change trend of each feature data within a preset time period are calculated. Based on the combination of the correlation and the change trend, the first sequence data or the second sequence data is divided into three types, each type corresponding to a target stage.

[0014] In combination with the first aspect, in the second implementation method of the first aspect of the present application, the target stage includes an initial stage, a mid-stage and a late stage, and the change trend of the characteristic data in each target stage corresponds to a different oxidation reaction rate range. The initial stage is a stage in which the oxidation reaction rate is less than a first threshold, the mid-stage is a stage in which the oxidation reaction rate is greater than or equal to the first threshold and less than a second threshold, and the late stage is a stage in which the oxidation reaction rate is greater than or equal to the second threshold, and the first sequence data and the second sequence data of each target stage are marked.

[0015] In conjunction with the first aspect, in a third implementation of the first aspect of the present application, establishing a corresponding prediction model for the initial stage includes:

[0016] The prediction model in the initial stage is a linear model, the first feature data is used as the input feature of the linear model, and the second feature data is used as the output feature of the linear model. The linear model is: Among them, Y i,j is the jth output feature of the i-th target stage, X k is the kth input feature, ω k For the input feature X k With the output feature Y i,j , α is the linear coupling coefficient of the output feature of the previous target stage to the output feature of the current target stage, Y i-1,j is the j-th output feature of the i-1-th target stage, b j is the bias term of the linear model, m is the total number of the input features, and multiple oxidation product indicators are weighted and combined to obtain the antioxidant index in the initial stage.

[0017] In combination with the first aspect, in a fourth implementation of the first aspect of the present application, a corresponding prediction model is established for the mid-term stage, including:

[0018] The prediction model for the mid-term stage is a neural network model, which takes the output features of the previous target stage and the first sequence data of the current mid-term stage as input features of the neural network model. The neural network model includes a first channel and a second channel, which are respectively used to receive the output features of the previous target stage and the first sequence data of the current mid-term stage. The neural network model outputs the predicted value of the oxidation product concentration, and calculates the antioxidant index based on the predicted concentration value. If the antioxidant index is less than the fourth threshold, this time is defined as the first time, and a first maintenance cycle is generated for the mid-term stage at the first time. During the first maintenance cycle, the knitting oil circulation system is notified to reapply the knitting machinery equipment.

[0019] In combination with the first aspect, in a fifth implementation of the first aspect of the present application, establishing a corresponding prediction model for the later stage includes:

[0020] In the later stage, an average difference between each first feature data in the first sequence data and all surrounding neighbor feature data is calculated; if the average difference is greater than a fifth threshold, the corresponding first feature data is marked as abnormal data, the abnormal data being abnormal operation data caused by oxidation of a non-knitting oil, and the abnormal data is removed from the first sequence data in the later stage;

[0021] The later stage is a deep learning model, which uses the remaining first sequence data and the output features of the mid-stage as input features of the deep learning model. The deep learning model outputs the oxidation rate of the knitting oil in the later stage, and obtains the antioxidant index of the knitting oil based on the oxidation rate. If the antioxidant index is less than a sixth threshold, the knitting oil circulation system is notified to replace the knitting oil within the second maintenance cycle.

[0022] In combination with the first aspect, in a sixth implementation of the first aspect of the present application, generating a first maintenance cycle for the mid-term stage includes:

[0023] Obtain the downward trend of the antioxidant index in the mid-term stage, determine the minimum safety value of the antioxidant index, perform multi-step prediction on the antioxidant prediction value based on the neural network model, generate a prediction sequence at multiple time points in the future, obtain the time corresponding to the antioxidant prediction value when it first falls below the minimum safety value in the prediction sequence, and define it as the second time. The first maintenance period is from the first time to the second time.

[0024] In combination with the first aspect, in a seventh implementation of the first aspect of the present application, notifying the knitting oil circulation system to replace the knitting oil within the second maintenance cycle includes:

[0025] The decrease rate of the antioxidant index per unit time in the later stage, as well as the equipment wear coupling coefficient and the emergency response time are obtained; based on the decrease rate, the average deviation degree of the oxidation state in the later stage from the initial oxidation state is obtained; based on the wear coupling coefficient, the effective life of the knitting machinery is predicted; and a risk function is established based on the average deviation degree and the effective life. The risk function is: Where D is the average deviation degree, T0 is the effective life, μ, is the weight coefficient, T1 is the safety threshold, D maxTo maximize the degree of deviation, a utility function is established based on the risk function. The utility function is: U(t) = -γR(t) - (1-γ)C(t), where γ is the risk preference coefficient and C(t) is the maintenance cost. The optimal maintenance time is obtained by maximizing the utility function, which is defined as the second maintenance cycle.

[0026] In a second aspect, the present application provides an intelligent evaluation system for the antioxidant properties of knitting oils, the system comprising:

[0027] A configuration module is configured to install a plurality of first sensors and second sensors, wherein the first sensors collect historical operating parameters of the knitting machine within a preset time, which are defined as first sequence data, and the second sensors collect historical oxidation products of the knitting oil within the preset time, which are defined as second sequence data;

[0028] a prediction module, configured to divide the oxidation process of the knitting oil into a plurality of target stages based on the first sequence data and the second sequence data, establish a corresponding prediction model for each target stage, wherein the prediction model is configured to predict the antioxidant performance of the knitting oil on the knitting machine equipment within the target stage and output an antioxidant index;

[0029] an analysis module, configured to obtain current operating parameters of the knitting machine during a real-time production process, locate the current oxidation stage of the knitting oil based on the current operating parameters, and output a corresponding antioxidant index based on a corresponding prediction model;

[0030] A maintenance module is configured to generate an adaptive maintenance cycle if the current oxidation stage is a middle stage or a late stage, and optimize the use of the knitting oil within the adaptive maintenance cycle.

[0031] A third aspect of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores instructions that, when executed on a computer, enable the computer to execute the above-mentioned intelligent evaluation method for the antioxidant properties of a knitting oil.

[0032] Compared with the prior art, the beneficial effects of the present invention are at least as follows:

[0033] By setting up a first sensor and a second sensor, comprehensive status data of knitting machinery and oil are collected to ensure data integrity and accuracy. Based on the correlation and change trend of the first and second sequence data, the oxidation process of the knitting oil is divided into an initial stage, a mid-stage, and a late stage. A linear model is established for the initial stage to predict the antioxidant index. A neural network model is established for the mid-stage to predict the concentration of oxidation products and calculate the antioxidant index. A deep learning model is established for the late stage to predict the oxidation rate and calculate the antioxidant index. Through quantitative standards, accurate division of the oxidation stages is achieved. According to the characteristics of different stages, appropriate prediction models are selected to improve the accuracy and reliability of the prediction. The current operating parameters of the knitting machinery in the real-time production process are obtained. Based on the current operating parameters, the current oxidation stage of the knitting oil is located, achieving real-time monitoring of the knitting oil oxidation process and timely obtaining the antioxidant index. It is determined whether the current oxidation stage is the mid-stage or the late stage. The mid-stage requires reapplying knitting oil during the first maintenance cycle, while the late stage requires replacing the oil during the second maintenance cycle. Through regular maintenance inspections and oil replacement, potential problems can be discovered and resolved in a timely manner, equipment failures can be prevented, and equipment reliability and stability can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0035] Figure 1 This is a schematic diagram of an embodiment of a method for intelligently evaluating the antioxidant properties of a knitting oil according to an embodiment of the present application;

[0036] Figure 2 Schematic diagram of analysis of all target stages in the embodiments of this application;

[0037] Figure 3 This is the generation process of the antioxidant index for each target stage in the embodiments of this application;

[0038] Figure 4 This is a schematic diagram of an embodiment of an intelligent evaluation system for the antioxidant properties of knitting oils in an embodiment of the present application. DETAILED DESCRIPTION

[0039] The embodiments of the present application provide a feed formula customization method, device, equipment and medium based on artificial intelligence. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of this application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0040] For ease of understanding, the specific process of the embodiment of the present application is described below. Figure 1 In the embodiment of the present application, an embodiment of an intelligent evaluation method for the antioxidant properties of a knitting oil comprises:

[0041] Step S1, installing multiple first sensors and second sensors, the first sensor collects historical operating parameters of the knitting machinery within a preset time, which is defined as the first sequence data, and the second sensor is used to collect historical oxidation products of the knitting oil within the preset time, which is defined as the second sequence data.

[0042] Specifically, taking the computerized flat knitting machine as an example, the computerized flat knitting machine is a common knitting equipment used to produce various knitted fabrics. Its main function is to weave yarns through the rotation of the needle cylinder and the movement of the needle bed.

[0043] The key components of a computerized flat knitting machine include a needle cylinder, a yarn feeding device, a pulling device, and a transmission system. A speed sensor and a position sensor can be installed on the needle cylinder, a tension sensor and a speed sensor can be installed on the yarn feeding device, a force sensor and a speed sensor can be installed on the pulling device, and a current sensor, a voltage sensor, and a vibration sensor can be installed in the transmission system. The historical operating parameters (such as speed, position, tension, speed, vibration, etc.) collected from the sensors are arranged in chronological order to form a time series data set, which is defined as the first sequence data. This data set contains the operating status and changes of the knitting machinery within a preset time.

[0044] Knitting oil degrades due to oxidation during use, affecting the performance of knitting machinery and fabric quality. To monitor the oil's quality, its historical oxidation products need to be monitored. Common oxidation products include the content of acidic substances in the oil (replaced by acid value) and peroxide content. For these oxidation products, a secondary sensor type, such as an acid value sensor or a peroxide sensor, is selected to monitor the content. The secondary sensor is installed at key locations in the oil circulation system, such as the oil tank, oil pipeline, or oil pump outlet, to ensure real-time monitoring of the oil's quality. The historical oxidation products collected by the second sensor are arranged in chronological order to form a time series dataset, defined as the second series data, which contains the quality changes of the knitting oil over a preset time period.

[0045] Step S2: Based on the first sequence data and the second sequence data, the oxidation process of the knitting oil is divided into multiple target stages, and a corresponding prediction model is established for each target stage. The prediction model is used to predict the antioxidant performance of the knitting oil on the knitting machinery and equipment within the target stage and output the antioxidant index.

[0046] Specifically, based on the changing trends of the first sequence data (equipment operating parameters) and the second sequence data (oil oxidation products), the oxidation process of knitting oil is divided into the following three stages: the initial stage (oil performance is stable and oxidation products increase slowly), the middle stage (oil performance begins to decline and oxidation products increase rapidly) and the late stage (oil performance deteriorates significantly and oxidation products increase sharply).

[0047] like Figure 2 As shown in the figure, it is an analysis diagram of all target stages. For each target stage, a suitable prediction model is selected. For example, a linear regression model can be selected and trained in the early stage, a machine learning model (such as a neural network model) can be selected in the mid-term stage, and a deep learning model (such as an LSTM model) can be selected in the late stage. Each prediction model predicts the peroxide content in this stage and generates an antioxidant index, where the higher the acid value, the lower the antioxidant index, and the higher the peroxide value, the lower the antioxidant index.

[0048] Step S3: obtaining the current operating parameters of the knitting machinery in the real-time production process, locating the current oxidation stage of the knitting oil based on the current operating parameters, and outputting the corresponding antioxidant index based on the corresponding prediction model.

[0049] Specifically, the current operating parameters of the knitting machinery and equipment and the current oxidation products of the knitting oil in the actual production process, for example (within 5 hours), are obtained, and the correlation and change trend between the current operating parameters and the current oxidation products are analyzed. The data are compared with the data of the three target stages (initial stage, middle stage and late stage). The similarity between the real-time data and the historical data of each target stage is calculated using a similarity measurement method (such as Euclidean distance, cosine similarity, dynamic time warping DTW), the real-time data is compared with the historical data of each target stage, the stage with the highest similarity is selected as the current oxidation stage, and the antioxidant index is predicted based on the prediction model of the current oxidation stage.

[0050] Step S4: If the current oxidation stage is the middle stage or the late stage, an adaptive maintenance cycle is generated, and the use of the knitting oil is optimized within the adaptive maintenance cycle.

[0051] Specifically, if the current phase is mid-term or late-term, an adaptive maintenance cycle is triggered. The specific cycle setting process is described later. During the cycle, an oil replenishment plan is developed based on oil usage and oxidation levels. For example, in the mid-term, fresh oil can be replenished based on oil consumption. In the late-term, an oil replacement plan is developed based on oil degradation. This adaptive maintenance cycle allows for precise maintenance and care based on actual oil usage and degradation, avoiding unnecessary oil changes and extending the oil's lifespan.

[0052] By setting up a first sensor and a second sensor, comprehensive status data of knitting machinery and oil are collected to ensure data integrity and accuracy. Based on the correlation and change trend of the first and second sequence data, the oxidation process of the knitting oil is divided into an initial stage, a mid-stage, and a late stage. A linear model is established for the initial stage to predict the antioxidant index. A neural network model is established for the mid-stage to predict the concentration of oxidation products and calculate the antioxidant index. A deep learning model is established for the late stage to predict the oxidation rate and calculate the antioxidant index. Through quantitative standards, accurate division of the oxidation stages is achieved. According to the characteristics of different stages, appropriate prediction models are selected to improve the accuracy and reliability of the prediction. The current operating parameters of the knitting machinery in the real-time production process are obtained. Based on the current operating parameters, the current oxidation stage of the knitting oil is located, achieving real-time monitoring of the knitting oil oxidation process and timely obtaining the antioxidant index. It is determined whether the current oxidation stage is the mid-stage or the late stage. The mid-stage requires reapplying knitting oil during the first maintenance cycle, while the late stage requires replacing the oil during the second maintenance cycle. Through regular maintenance inspections and oil replacement, potential problems can be discovered and resolved in a timely manner, equipment failures can be prevented, and equipment reliability and stability can be improved.

[0053] In a specific embodiment, dividing the oxidation process of the knitting oil into multiple target stages specifically includes the following steps:

[0054] First characteristic data and second characteristic data are extracted from the first sequence data and the second sequence data respectively, the first characteristic data including the time series characteristics, frequency domain characteristics, peak characteristics and extreme value characteristics of historical operating parameters, and the second characteristic data including the concentration change characteristics, cumulative characteristics and fluctuation characteristics of historical oxidation products. The correlation between the first characteristic data and the second characteristic data and the change trend of each characteristic data within a preset time period are calculated. Based on the combination of the correlation and the change trend, the first sequence data or the second sequence data is divided into three types, and each type corresponds to a target stage.

[0055] Specifically, the first and second series of data are time-aligned to ensure that the sensor data correspond to the same time point. First and second feature data are then extracted from the first and second series of data, respectively. Time series features in the first feature data include trend characteristics and rate of change. Trend characteristics involve calculating moving averages, exponential smoothing, and other methods to extract parameter trend information. Edge rates include temperature and pressure rate of change. Frequency domain features are primarily acquired by performing a Fourier transform on the time series data to extract the primary frequency components, calculating the energy distribution of different frequency components, and identifying the periodicity of parameter changes. Peak features include the maximum and minimum values, or kurtosis of a parameter. Extreme value features include the number of times a parameter reaches an extreme value within a preset time period and the time at which the extreme value occurs. Second feature data includes concentration change features such as the rate of change of acidity and peroxide. Cumulative features include the total amount of oxidation products accumulated within a preset time period. Fluctuation features include the fluctuation range of oxidation products.

[0056] Use correlation analysis (such as the Pearson correlation coefficient) to calculate the degree of association between the first feature data and the second feature data, constructing a correlation matrix to display the strength of associations between different features. For example, the correlation between temperature and acid value, the correlation between pressure and peroxide value, etc. Based on the characteristic information, trend analysis is performed on the first and second sequence data to identify their changing trends. For example, linear regression, exponential smoothing, or ARIMA models are used to fit data trends. For example, an upward trend: the concentration of a parameter or oxidation product increases over time; a downward trend: the concentration of a parameter or oxidation product decreases over time; and a stable trend: the concentration of a parameter or oxidation product remains relatively stable. The correlation between the first sequence data and the second sequence data is combined with their respective change trends. For example, high correlation + rising trend: indicates that there is a strong correlation between the operating parameters and the oxidation products, and the concentration of the oxidation products is increasing. When the temperature rises to a certain threshold, there is a strong correlation between the temperature and the acid value. This is because the increase in temperature accelerates the oxidation reaction of the oil. According to the combination of the above correlation and change trend, the first sequence data and the second sequence data are divided into three stages of data, namely the initial stage, the middle stage and the late stage. The data combination type of the initial stage is: low correlation + stable trend or low correlation + slowly rising trend; the data combination type of the middle stage is medium correlation + accelerated rising trend; the combination type feature of the late stage is high correlation + sharply rising trend.

[0057] In a specific embodiment, the target stage includes an initial stage, a middle stage and a late stage. The change trend of the characteristic data in each target stage corresponds to a different oxidation reaction rate range. The initial stage is a stage in which the oxidation reaction rate is less than a first threshold, the middle stage is a stage in which the oxidation reaction rate is greater than or equal to the first threshold and less than a second threshold, and the late stage is a stage in which the oxidation reaction rate is greater than or equal to the second threshold. The first sequence data and the second sequence data of each target stage are marked.

[0058] Specifically, in the initial stage, the oxidation reaction of the oil agent has just begun, the concentration of oxidation products changes slowly, the equipment operating parameters are relatively stable, and the changing trend of the characteristic data is stable or slowly rising; in the mid-term stage: as time goes by, the oxidation reaction rate accelerates, the concentration of oxidation products increases rapidly, the equipment operating parameters begin to fluctuate, and the changing trend of the characteristic data accelerates; in the late stage: the oxidation reaction rate increases sharply, the concentration of oxidation products rises sharply, the equipment operating parameters show obvious abnormalities, and the changing trend of the characteristic data rises rapidly. The operating parameters and oxidation product data of each stage are marked.

[0059] In one embodiment, establishing a corresponding prediction model for the initial stage specifically includes the following steps:

[0060] The prediction model in the initial stage is a linear model. The first feature data is used as the input feature of the linear model, and the second feature data is used as the output feature of the linear model. The linear model is: Among them, Y i,j is the jth output feature of the i-th target stage, X k is the kth input feature, ω k is the input feature X k With the output feature Y i,j , α is the linear coupling coefficient of the output feature of the previous target stage to the output feature of the current target stage, Y i-1,j is the j-th output feature of the i-1-th target stage, b j is the bias term of the linear model, m is the total number of input features, and multiple oxidation product indicators are weightedly combined to obtain the antioxidant index in the initial stage.

[0061] Specifically, if Figure 3 As shown in the figure, the generation process of the antioxidant index for each target stage is as follows. Since the oxidation reaction rate in the initial stage is slow and the change trend is relatively stable, a linear model is selected to describe the oxidation process in the initial stage. The linear model formula is expressed by the weight coefficient ω. k For different input features X k Weighted to reflect the importance of each feature to the output feature, αY i-1,j For historical output feature coupling, the output features of the previous target stage are introduced into the current model to reflect the impact of historical data on the current prediction. The information of historical data is used to improve the prediction stability of the model. For example, the concentration of oxidation products at the previous time point can be used as an important reference for the prediction at the current time point, improving the continuity and accuracy of the prediction. j The bias term is used to adjust the output of the model to make it more consistent with the actual situation. The real-time operating parameters and oxidation product data are input into the trained linear model, and the linear model is used to predict the current antioxidant index. The linear model has a fast calculation speed and can process data in real time to realize real-time monitoring of the knitting oil oxidation process.

[0062] In a specific embodiment, establishing a corresponding prediction model for the mid-term stage specifically includes the following steps:

[0063] The prediction model for the mid-term stage is a neural network model, which takes the output features of the previous target stage and the first sequence data of the current mid-term stage as input features of the neural network model. The neural network model includes a first channel and a second channel, which are respectively used to receive the output features of the previous target stage and the first sequence data of the current mid-term stage. The neural network model outputs a predicted value of the oxidation product concentration and calculates the antioxidant index based on the concentration prediction value. If the antioxidant index is less than the fourth threshold, this time is defined as the first time, and the first maintenance cycle is generated for the mid-term stage at the first time. During the first maintenance cycle, the knitting oil circulation system is notified to reapply the knitting machinery equipment.

[0064] Specifically, a neural network model is selected as the prediction model for the mid-term phase because the oxidation reaction rate in the mid-term phase is faster and the change trend is complex, making it suitable for prediction using a neural network model. Output feature data, such as the antioxidant index and oxidation product concentration, are obtained from the prediction model of the previous target phase, for example, the initial phase. Operating parameter data for the current mid-term phase, such as temperature 35°C, pressure 200Pa, and speed 1500RPM, are obtained from the first sensor. The model processes the output features of the initial phase and the first sequence data of the current mid-term phase through the first channel and the second channel, respectively. The neural network model outputs a predicted oxidation product concentration, for example, a predicted acid value of 3.5mgKOH / g. According to the formula: antioxidant index = 1-(predicted oxidation product concentration / maximum oxidation product concentration), the antioxidant index is calculated to be 0.7. Assuming the fourth threshold is 0.9, since 0.7<0.9, a maintenance cycle is triggered. At the first time, the first maintenance cycle is generated, notifying the knitting oil circulation system to re-apply oil to the equipment. At this time, the performance of the oil has significantly declined, but has not yet completely deteriorated. Reapplying fresh oil can restore lubrication performance and slow oil degradation. The neural network model's rapid computational speed allows for real-time data processing, enabling real-time monitoring of the knitting oil's oxidation process. This allows for early detection of the oil's antioxidant capacity falling below a preset threshold, allowing for early intervention and replenishment, improving efficiency and reducing operating costs.

[0065] In one embodiment, establishing a corresponding prediction model for the later stage specifically includes the following steps:

[0066] In the later stage, the average difference between each first feature data in the first sequence data and all neighboring feature data is calculated. If the average difference is greater than a fifth threshold, the corresponding first feature data is marked as abnormal data. The abnormal data is abnormal operation data caused by oxidation of non-knitting oil, and the abnormal data is removed from the first sequence data in the later stage.

[0067] The later stage is a deep learning model, which uses the remaining first sequence data and the output features of the mid-stage as the input features of the deep learning model. The deep learning model outputs the oxidation rate of the knitting oil in the later stage, and obtains the antioxidant index of the knitting oil based on the oxidation rate. If the antioxidant index is less than the sixth threshold, the knitting oil circulation system is notified to replace the knitting oil within the second maintenance cycle.

[0068] Specifically, for each first feature data (e.g., temperature, pressure, rotation speed, etc.) in the first sequence data of the later stage, the average difference between it and the feature data of the surrounding neighbors is calculated. The average difference can be calculated using the following formula: X t is the feature data at the current time point, X t-i is the characteristic data of the previous i time points, N is the number of neighbors, and the "average distance" refers to the average of the numerical differences between the current data point and its surrounding neighboring data points in the time series data. Specifically, it is used to measure the similarity between the current data point and the adjacent data points. Some abnormal data points may not conform to the overall trend, but show significant differences in the local range. By calculating the average distance, these local anomalies can be effectively identified. For example, in a normally operating device, a data point with a sudden temperature rise suddenly appears. Its average distance from the surrounding data will be large, and it will be easily identified as an anomaly. In addition, during the device startup phase or when the load changes, the data fluctuates greatly. The average distance method can dynamically adjust the judgment criteria to avoid misjudging normal fluctuations as anomalies.

[0069] The late-stage model uses a deep learning model (such as LSTM), which can handle more complex time series data and long-term dependencies compared to the neural network model in the mid-stage, and is suitable for the rapidly changing oxidation process in the late stage. The first sequence data after removing anomalies and the output features of the mid-stage are used as the input of the deep learning model, reflecting the impact of the mid-stage on the late stage. The deep model predicts the oxidation rate and calculates the antioxidant index based on the oxidation rate. The formula is as follows: Antioxidant Index = e -ρ*氧化速度 If the calculated antioxidant index falls below the set threshold, a maintenance strategy is triggered. In the later stages of the process, the additives in the knitting oil have been largely consumed, and excessive accumulation of oxidation products (such as acid value and peroxide value) has occurred, leading to fundamental changes in the oil's properties. For example, an excessively high acid value can corrode the equipment, while an excessively high peroxide value can cause the oil to decompose, producing precipitation and colloids. At this point, the oil's performance cannot be restored by simply reapplying it, and replacement is necessary to ensure normal operation and extend the equipment's service life.

[0070] In a specific embodiment, generating a first maintenance cycle for the mid-term phase specifically includes the following steps:

[0071] Obtain the downward trend of the antioxidant index in the mid-term stage, determine the minimum safe value of the antioxidant index, perform multi-step prediction of the antioxidant prediction value based on the neural network model, generate a prediction sequence at multiple time points in the future, and obtain the time corresponding to the antioxidant prediction value when it first falls below the minimum safe value in the prediction sequence, which is defined as the second time. The first maintenance cycle is from the first time to the second time.

[0072] Specifically, assuming that the current time is 10:00 on March 10, 2025, the downward trend of the antioxidant index is obtained, antioxidant index = 1.0-0.05×t, the current antioxidant index is assumed to be 0.8, the minimum safety value is determined to be 0.5, the prediction sequence is: [0.8, 0.75, 0.7, 0.65, 0.6, 0.55, 0.5, 0.45, 0.4, 0.35], find the time point when it is first lower than 0.5: the 7th time point (t=6), then the first maintenance cycle is generated: the first time is 10:00 on March 10, 2025, the second time is 16:00 on March 10, 2025, the first maintenance cycle: from 10:00 on March 10, 2025 to 16:00 on March 10, 2025, then inform the operator that the oil needs to be reapplied before 16:00 on March 10, 2025. Based on multi-step prediction results, maintenance time is accurately determined to avoid maintenance being performed too early or too late. By calculating the first maintenance cycle, oil re-application is carried out in a timely manner to avoid damage to the equipment caused by oil deterioration and improve equipment reliability and stability.

[0073] In a specific embodiment, notifying the knitting oil circulation system to replace the knitting oil in the second maintenance cycle specifically includes the following steps:

[0074] The decrease rate of the antioxidant index per unit time in the later stage, as well as the equipment wear coupling coefficient and emergency response time, are obtained. Based on the decrease rate, the average deviation degree between the oxidation state in the later stage and the initial oxidation state is obtained. The effective life of the knitting machinery is predicted based on the wear coupling coefficient. A risk function is established based on the average deviation degree and the effective life. The risk function is: Where D is the average deviation degree, T0 is the effective life, μ, is the weight coefficient, T1 is the safety threshold, D max To maximize the degree of deviation, a utility function is established based on the risk function. The utility function is: U(t) = -γR(t)-(1-γ)C(t), where γ is the risk preference coefficient and C(t) is the maintenance cost. The optimal maintenance time is obtained by maximizing the utility function, which is defined as the second maintenance cycle.

[0075] Specifically, the decrease rate of the antioxidant index per unit time in the later stage, as well as the wear coupling coefficient and emergency response time of the knitting machinery are obtained. The wear coupling coefficient reflects the impact of the degree of equipment wear on the performance of the oil, and the emergency response time reflects the response speed when the equipment fails. The degree of oxidation state deviation is calculated according to the above formula, which is: Among them, D is the average deviation degree, T is the time period antioxidant index t is the antioxidant index at time point t in the later stage, and the initial antioxidant index t The antioxidant index at time point t in the initial stage, assuming that the initial antioxidant index is 1.0, the later antioxidant index is 0.5, and the time period is 10 hours, is brought into the calculation to obtain D = 0.5. The maximum deviation degree is the deviation degree of the equipment under the maximum wear condition. Assuming the maximum deviation degree is 1.0, the wear coupling coefficient is 0.2, and the effective life is 1.0 / 0.2 = 5 hours. By establishing the risk function, assuming μ = 0.6, D max =1.0, T1=2, T0=5, and various weighting factors are set and brought into the calculation to obtain a risk value of 0.46. This is then incorporated into the utility function, and the time point that maximizes the utility function is calculated and defined as the second maintenance cycle. Based on the antioxidant index, equipment wear coupling coefficient, and effective lifespan, maintenance time is accurately determined, taking into account emergency response time to ensure timely handling of equipment failures, reduce downtime, and promptly replace lubricants to prevent damage to equipment caused by lubricant degradation, thereby improving equipment reliability and stability.

[0076] The above describes a knitting oil antioxidant intelligent evaluation method in the embodiment of the present application. The following describes a knitting oil antioxidant intelligent evaluation system in the embodiment of the present application. Figure 4 In one embodiment of the present application, a knitting oil antioxidant intelligent evaluation system includes:

[0077] A configuration module is configured to install a plurality of first sensors and second sensors, wherein the first sensors collect historical operating parameters of the knitting machinery within a preset time, which are defined as first sequence data, and the second sensors collect historical oxidation products of the knitting oil within a preset time, which are defined as second sequence data;

[0078] a prediction module for dividing the oxidation process of the knitting oil into a plurality of target stages based on the first sequence data and the second sequence data, establishing a corresponding prediction model for each target stage, the prediction model being used to predict the antioxidant performance of the knitting oil on the knitting machinery and equipment within the target stage and outputting an antioxidant index;

[0079] An analysis module is used to obtain the current operating parameters of the knitting machinery in the real-time production process, locate the current oxidation stage of the knitting oil based on the current operating parameters, and output the corresponding antioxidant index based on the corresponding prediction model;

[0080] The maintenance module is used to generate an adaptive maintenance cycle if the current oxidation stage is the middle stage or the late stage, and optimize the use of the knitting oil within the adaptive maintenance cycle.

[0081] The present application also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. The computer-readable storage medium stores instructions, which, when executed on a computer, cause the computer to execute the steps of a method for intelligently evaluating the antioxidant properties of a knitting oil.

[0082] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0083] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, and other media that can store program code.

[0084] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. An intelligent evaluation method for the antioxidant properties of knitting oils, characterized in that: The method comprises: Step S1: installing a plurality of first sensors and second sensors, wherein the first sensors collect historical operating parameters of the knitting machine within a preset time, which is defined as first sequence data, and the second sensors collect historical oxidation products of the knitting oil within the preset time, which is defined as second sequence data; Step S2: dividing the oxidation process of the knitting oil into multiple target stages based on the first sequence data and the second sequence data, establishing a corresponding prediction model for each target stage, the prediction model being used to predict the antioxidant performance of the knitting oil on the knitting machine equipment within the target stage and outputting an antioxidant index; Step S3: obtaining current operating parameters of the knitting machine during real-time production, locating the current oxidation stage of the knitting oil based on the current operating parameters, and outputting a corresponding antioxidant index based on a corresponding prediction model; Step S4: If the current oxidation stage is the middle stage or the late stage, an adaptive maintenance cycle is generated, and the use of the knitting oil is optimized within the adaptive maintenance cycle.

2. The method according to claim 1, characterized in that The oxidation process of the knitting oil is divided into multiple target stages including: First feature data and second feature data are extracted from the first sequence data and the second sequence data, respectively. The first feature data includes time series features, frequency domain features, peak features, and extreme value features of historical operating parameters. The second feature data includes concentration change features, accumulation features, and fluctuation features of historical oxidation products. The correlation between the first feature data and the second feature data and the change trend of each feature data within a preset time period are calculated. Based on the combination of the correlation and the change trend, the first sequence data or the second sequence data is divided into three types, each type corresponding to a target stage.

3. The method according to claim 2, characterized in that The target stage includes an initial stage, a mid-stage, and a late stage. The changing trend of the characteristic data in each target stage corresponds to a different oxidation reaction rate range. The initial stage is a stage in which the oxidation reaction rate is less than a first threshold, the mid-stage is a stage in which the oxidation reaction rate is greater than or equal to the first threshold and less than a second threshold, and the late stage is a stage in which the oxidation reaction rate is greater than or equal to the second threshold. The first sequence data and the second sequence data of each target stage are marked.

4. The method according to claim 3, characterized in that The corresponding prediction models for the initial stage include: The prediction model in the initial stage is a linear model, the first feature data is used as the input feature of the linear model, and the second feature data is used as the output feature of the linear model. The linear model is: Among them, Y i,j is the jth output feature of the i-th target stage, X k is the kth input feature, ω k For the input feature X k With the output feature Y i,j , α is the linear coupling coefficient of the output feature of the previous target stage to the output feature of the current target stage, Y i-1,j is the j-th output feature of the i-1-th target stage, b j is the bias term of the linear model, m is the total number of the input features, and multiple oxidation product indicators are weighted and combined to obtain the antioxidant index in the initial stage.

5. The method according to claim 4, characterized in that The corresponding prediction model for the mid-term stage includes: The prediction model for the mid-term stage is a neural network model, which takes the output features of the previous target stage and the first sequence data of the current mid-term stage as input features of the neural network model. The neural network model includes a first channel and a second channel, which are respectively used to receive the output features of the previous target stage and the first sequence data of the current mid-term stage. The neural network model outputs the predicted value of the oxidation product concentration, and calculates the antioxidant index based on the predicted concentration value. If the antioxidant index is less than the fourth threshold, this time is defined as the first time, and a first maintenance cycle is generated for the mid-term stage at the first time. During the first maintenance cycle, the knitting oil circulation system is notified to reapply the knitting machinery equipment.

6. The method according to claim 5, characterized in that Building corresponding prediction models for the later stages includes: In the later stage, an average difference between each first feature data in the first sequence data and all surrounding neighbor feature data is calculated; if the average difference is greater than a fifth threshold, the corresponding first feature data is marked as abnormal data, the abnormal data being abnormal operation data caused by oxidation of a non-knitting oil, and the abnormal data is removed from the first sequence data in the later stage; The later stage is a deep learning model, which uses the remaining first sequence data and the output features of the mid-stage as input features of the deep learning model. The deep learning model outputs the oxidation rate of the knitting oil in the later stage, and obtains the antioxidant index of the knitting oil based on the oxidation rate. If the antioxidant index is less than a sixth threshold, the knitting oil circulation system is notified to replace the knitting oil within the second maintenance cycle.

7. The method according to claim 5, characterized in that Generating a first maintenance cycle for the mid-term phase, including: Obtain the downward trend of the antioxidant index in the mid-term stage, determine the minimum safety value of the antioxidant index, perform multi-step prediction on the antioxidant prediction value based on the neural network model, generate a prediction sequence at multiple time points in the future, obtain the time corresponding to the antioxidant prediction value when it first falls below the minimum safety value in the prediction sequence, and define it as the second time. The first maintenance period is from the first time to the second time.

8. The method according to claim 6, characterized in that Then, the knitting oil circulation system is notified to replace the knitting oil in the second maintenance cycle, including: The decrease rate of the antioxidant index per unit time in the later stage, as well as the equipment wear coupling coefficient and the emergency response time are obtained; based on the decrease rate, the average deviation degree of the oxidation state in the later stage from the initial oxidation state is obtained; based on the wear coupling coefficient, the effective life of the knitting machinery is predicted; and a risk function is established based on the average deviation degree and the effective life. The risk function is: Where D is the average deviation degree, T0 is the effective life, μ, is the weight coefficient, T1 is the safety threshold, D max To maximize the degree of deviation, a utility function is established based on the risk function. The utility function is: U(t) = -γR(t) - (1-γ)C(t), where γ is the risk preference coefficient and C(t) is the maintenance cost. The optimal maintenance time is obtained by maximizing the utility function, which is defined as the second maintenance cycle.

9. An intelligent evaluation system for the antioxidant properties of knitting oils, for implementing the method according to any one of claims 1 to 8, characterized in that: The system comprises: A configuration module is configured to install a plurality of first sensors and second sensors, wherein the first sensors collect historical operating parameters of the knitting machine within a preset time, which are defined as first sequence data, and the second sensors collect historical oxidation products of the knitting oil within the preset time, which are defined as second sequence data; a prediction module, configured to divide the oxidation process of the knitting oil into a plurality of target stages based on the first sequence data and the second sequence data, establish a corresponding prediction model for each target stage, wherein the prediction model is configured to predict the antioxidant performance of the knitting oil on the knitting machine equipment within the target stage and output an antioxidant index; an analysis module, configured to obtain current operating parameters of the knitting machine during a real-time production process, locate the current oxidation stage of the knitting oil based on the current operating parameters, and output a corresponding antioxidant index based on a corresponding prediction model; A maintenance module is configured to generate an adaptive maintenance cycle if the current oxidation stage is a middle stage or a late stage, and optimize the use of the knitting oil within the adaptive maintenance cycle.

10. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instructions are executed by a processor, the method according to any one of claims 1 to 8 is implemented.

Images