Dynamic feedback-based centrifuge parameter self-adjusting system and method for nodular cast iron pipe

Abstract

The application discloses a nodular cast iron pipe centrifuge parameter self-adjusting system and method based on dynamic feedback, and belongs to the technical field of nodular cast iron pipe production optimization, and comprises a data acquisition module, an identification module, an adjusting module and an execution feedback module; the data acquisition module is used for collecting parameter actual values in real time, and the parameter actual values are pretreated and saved into a parameter database; the identification module is used for performing parameter deviation analysis on each parameter in the parameter database, identifying deviation parameters of the nodular cast iron pipe centrifuge, classifying and summarizing the deviation parameters, and identifying associated parameters affected by the deviation parameters and influence degrees; the adjusting module is used for calculating optimal running values and predicted target values of each parameter through a set target function and constraint conditions; and the execution feedback module is used for adjusting according to the optimal running values, continuously collecting feedback data, calculating actual target values and performing real-time feedback.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of nodular cast iron pipe production optimization, and relates to a nodular cast iron pipe centrifuge parameter self-adjusting system and method based on dynamic feedback. BACKGROUND

[0002] In the production process of nodular cast iron pipes, the centrifuge is a key equipment, and accurate control of its parameters has a decisive influence on pipe quality. In the operation of traditional centrifuges, parameters are mostly pre-set, and it is difficult to adapt to various dynamic changes in the production process, such as fluctuations in molten iron temperature, differences in raw material composition, and equipment component wear.

[0003] The existing Chinese patent with publication number CN109158569B discloses a water-cooled nodular cast pipe centrifuge and a control system thereof, which includes an input layer data acquisition system, an implicit layer data system, and an output layer data acquisition system. The data collected by the PLC is uploaded to the intelligent unit real-time database, the centrifuge production experience is adjusted in combination with the centrifuge operation experts, the centrifuge casting model is generated by using the machine deep learning algorithm to optimize the pipe casting parameters, the optimized centrifuge parameters are automatically written into the PLC, and the centrifuge speed, the package turning speed, the rotation speed, and the position control are controlled by the PLC.

[0004] Although the existing technology realizes the stability of nodular cast pipes under automatic control to ensure the uniformity of wall thickness, reduces the labor intensity of workers, reduces production costs, improves product quality, and improves production efficiency, it does not consider real-time feedback and rapid response of dynamic changes. For example, the molten iron temperature is prone to rapid fluctuations in the production process, which can have a very significant impact on the entire casting process in a short time. When the molten iron temperature rises rapidly, its fluidity will change, thereby affecting the distribution of the molten iron under the action of centrifugal force, and ultimately having a negative effect on the uniformity of the pipe wall thickness. Meanwhile, the comprehensive adjustment under the coupling action of multiple factors is not considered. In actual production scenarios, factors such as fluctuations in molten iron temperature and differences in raw material composition do not exist in isolation, but are intertwined and interact with each other. For example, fluctuations in molten iron temperature and differences in raw material composition will produce a coupling effect. When the molten iron temperature rises, if the carbon content in the raw material is relatively low, the solidification process of the molten iron may be affected in the casting process, causing changes in solidification time and solidification mode, and thereby having a more complex impact on the internal organizational structure and performance of the pipe. Therefore, the present application provides a nodular cast iron pipe centrifuge parameter self-adjusting system and method based on dynamic feedback. SUMMARY

[0005] In view of the deficiencies in the prior art, the purpose of the present application is to provide a nodular cast iron pipe centrifuge parameter self-adjusting system and method based on dynamic feedback, which classifies and summarizes deviated parameters and takes different processing measures through parameter deviation analysis and hierarchical judgment mechanism, considers the interaction of multiple factors, and also adjusts parameters according to real-time feedback, judges the adjustment effect according to the product qualification rate, re-evaluates the coupling relationship if it does not meet the expectation, and realizes real-time feedback and rapid response to dynamic changes and comprehensive adjustment under the coupling of multiple factors.

[0006] To achieve the above purpose, the present application provides the following technical solutions:

[0007] The nodular cast iron pipe centrifuge parameter self-adjusting system based on dynamic feedback comprises a data acquisition module, an identification module, an adjustment module and an execution feedback module.

[0008] The data acquisition module is used for real-time acquisition of parameter actual values, pre-processing of the parameter actual values and saving the parameter actual values to a constructed parameter database.

[0009] The identification module is used for parameter deviation analysis of each parameter in the parameter database, identification of deviated parameters of the nodular cast iron pipe centrifuge, hierarchical summary of the deviated parameters, and identification of associated parameters affected by the deviated parameters and the influence degree.

[0010] The adjustment module is used for calculation of the best running value and the predicted target value of each parameter through a set target function and constraint condition.

[0011] The execution feedback module is used for adjustment according to the best running value, continuous collection of feedback data, calculation of the actual target value and real-time feedback.

[0012] Further, the identification module comprises an abnormality detection unit and a coupling matching unit.

[0013] The abnormality detection unit is used for parameter deviation analysis according to a set analysis interval, dynamic judgment of whether each parameter deviates from the normal running track during the running process of the nodular cast iron pipe centrifuge, and hierarchical summary.

[0014] The coupling matching unit is used for analysis of the influence of the deviated parameters on other parameters, acquisition of associated parameters and calculation of the influence degree.

[0015] Further, the specific steps of parameter deviation analysis comprise:

[0016] The time window is set to , the parameter is calculated, the moving average value and the standard deviation in the preset time window are calculated, and the standard deviation multiple is set to whether the parameter has a slight deviation;

[0017] If , it is determined that the parameter has no slight deviation; if , it is determined that the parameter has a slight deviation; wherein is the actual value of the parameter at the first time point;

[0018] a trend slope of the parameter in the time window and a historical trend slope are calculated;

[0019] a deviation threshold is set as , a difference threshold is set as , and a trend difference value of the parameter is calculated to determine whether the parameter has a trend deviation;

[0020] If and , it is determined that the parameter has a trend deviation; if or , it is determined that the parameter has no trend deviation.

[0021] Further, after the parameter deviation analysis, each parameter is classified and summarized, and the specific rules of the classification and summary are as follows:

[0022] a parameter having both a slight deviation and a trend deviation is defined as a deviation parameter, and the deviation parameter is transmitted to the coupling matching unit;

[0023] a parameter having no slight deviation and no trend deviation is defined as a normal parameter, and the parameter deviation analysis is continuously performed;

[0024] a parameter having only a slight deviation or a trend deviation is defined as a potential parameter, and a potential analysis is continuously performed.

[0025] Further, the specific steps of the potential analysis include:

[0026] based on a sampling interval , the potential parameter is continuously subjected to the parameter deviation analysis;

[0027] ​If both the slight deviation and the trend deviation exist after each parameter deviation analysis, the potential parameter is updated as the deviation parameter, otherwise the number of deviations is recorded ;

[0028] After performing times of parameter deviation analysis, the frequency of deviation is calculated , and a frequency threshold is set to determine whether the potential parameter is abnormal;

[0029] If , the potential parameter is updated as the deviation parameter;

[0030] If , before the latest parameter deviation analysis, if there is no deviation for times in succession, the potential parameter is updated as the normal parameter, otherwise, it is kept as the potential parameter and continues to be analyzed according to the sampling interval.

[0031] Further, the specific steps of calculating the influence degree include:

[0032] Based on the parameter database, the Pearson correlation coefficient between any two parameters is calculated, and a parameter correlation matrix is constructed ;

[0033] In turn, each parameter is taken as the dependent variable and other parameters are taken as the independent variable to construct a preliminary model, and the preliminary model is evaluated and modified using process knowledge and expert experience to generate a coupling relationship model;

[0034] The correlation strength threshold is set to , and in the parameter correlation matrix, the matrix elements corresponding to the deviation parameter are traversed, and the parameters that satisfy are defined as associated parameters and saved to the influence parameter list, and the associated parameters are screened using the coupling relationship model to update the influence parameter list; wherein is the Pearson correlation coefficient between the deviation parameter and the parameter , , is the number of parameters, and ;

[0035] The change amount of the deviation parameter is substituted into the coupling relationship model to calculate the corresponding change amount of the associated parameters in the influence parameter list.

[0036] Further, the specific steps of calculating the optimal operation value include:

[0037] Defining the associated parameters in the list of deviation parameters and influence parameters as the parameters to be adjusted, and establishing a target function with the optimization goal of minimizing product defect rate ;

[0038] Setting constraints, including the value range of the parameters to be adjusted, the adjustable amount, and the associated constraints between the parameters to be adjusted;

[0039] Using gradient descent method to iteratively update the parameters to be adjusted until the target function converges to a stable value , and outputting the value of the parameters to be adjusted , and Defining the optimal operation value as the predicted target value, and Defining the predicted target value as the predicted target value.

[0040] Further, the specific steps of real-time feedback include:

[0041] Obtaining the total output of the ductile iron pipe after parameter adjustment and the product qualified amount , and calculating the actual target value ;

[0042] Setting the target difference threshold as , and calculating the predicted difference between the actual target value and the predicted target value ;

[0043] If , continue the parameter deviation analysis;

[0044] If , the actual effect does not meet the expectation, update the coupling relationship model, and re-determine the parameters to be adjusted and the optimal operation value according to the newly evaluated coupling relationship until the actual effect meets the expectation.

[0045] The parameter self-adjusting method of the ductile iron pipe centrifuge based on dynamic feedback includes:

[0046] Real-time collection of parameter actual values and saving to the constructed parameter database;

[0047] Performing parameter deviation analysis according to the set analysis interval, and performing hierarchical induction;

[0048] Analyzing the influence of the deviation parameters on other parameters, obtaining the associated parameters, and calculating the influence degree;

[0049] Calculating the optimal operation value and the predicted target value of each parameter through the set target function and constraints;

[0050] According to the optimal operation value, the feedback data is continuously collected, the actual target value is calculated, and real-time feedback is performed.

[0051] Advantages of the present application:

[0052] The parameter deviation analysis is performed by using the moving average method, trend analysis method, etc., a hierarchical judgment mechanism is introduced, the abnormal parameters are accurately identified, through the all-around monitoring and analysis of the parameters, the abnormality of the nodular cast iron pipe centrifuge in operation can be found in time, the basis for taking corresponding measures for different deviation degrees is provided, the accuracy and efficiency of production management are improved; the correlation analysis of the deviated parameters can be performed, the affected parameters and the influence degree can be determined, and then the optimal operation value is calculated by setting the target function and the constraint condition, which helps to get rid of the influence of parameter deviation, restore the optimal operation state, improve the product quality and production benefit, so as to realize the optimization adjustment; the feedback data is continuously collected after the adjustment, the actual and predicted target values are compared to judge the adjustment effect, and the real-time feedback mechanism can continuously optimize the adjustment strategy according to the actual production effect, so that the production process is always in a good state. BRIEF DESCRIPTION OF DRAWINGS

[0053] Figure 1 It is a structure diagram of the nodular cast iron pipe centrifuge parameter self-adjusting system based on dynamic feedback.

[0054] Figure 2 It is a flowchart of the parameter deviation analysis of the present application.

[0055] Figure 3 It is a flowchart of the calculation of the influence degree of the present application.

[0056] Figure 4 It is a flowchart of the real-time feedback of the present application.

[0057] Figure 5 It is a flowchart of the nodular cast iron pipe centrifuge parameter self-adjusting method based on dynamic feedback. DETAILED DESCRIPTION

[0058] The technical scheme of the present application will be described in detail below with reference to the drawings and specific embodiments, and it should be understood that the embodiments and specific features in the embodiments are detailed descriptions of the technical scheme of the present application, and are not limitations of the technical scheme of the present application, and the technical features in the embodiments and the embodiments can be combined with each other without conflict.

[0059] Embodiment 1

[0060] Reference Figures 1 to 4 As shown in the figure, the present embodiment introduces a nodular cast iron pipe centrifuge parameter self-adjusting system based on dynamic feedback, which comprises a data acquisition module, an identification module, an adjustment module, an execution feedback module and a communication module.

[0061] The data acquisition module is used to collect the actual values of various parameters affecting the quality of the ductile cast iron pipe in real time during the operation process through various high-precision sensors installed at key positions of the ductile cast iron pipe centrifuge, and at the same time, a timestamp is added to each actual value to accurately record the time when the data is collected, so that subsequent data analysis and quality control can be performed in chronological order, and the actual values of the parameters are preprocessed, and the preprocessed actual values of the parameters are saved to the constructed parameter database; wherein the preprocessing includes removing outliers, filtering and denoising, setting reasonable upper and lower limit ranges for each parameter, defining data points exceeding the range as outliers and automatically removing them to avoid misleading subsequent analysis, using Kalman filtering algorithm to remove interference signals and improve data quality, and the parameters affecting the quality of the ductile cast iron pipe include but are not limited to centrifuge speed, molten iron temperature, mold temperature, and cooling water flow;

[0062] The identification module is used to perform parameter deviation analysis on each parameter in the parameter database, identify the deviation parameters of the ductile cast iron pipe centrifuge, classify and summarize the deviation parameters, and identify the associated parameters affected by the deviation parameters and the influence degree;

[0063] The adjustment module is used to calculate the best operating value and the predicted target value of each parameter under the current production conditions by reasonably setting the target function and the constraint condition, so as to get rid of the adverse effects of parameter deviation and restore to the best operating state, thereby realizing the optimization adjustment of the system;

[0064] The execution feedback module is used to adjust each parameter of the ductile cast iron pipe centrifuge in real time according to the calculated best operating value, and continuously collect feedback data after adjusting the parameters, including the real-time values of each parameter after adjustment, the quality detection data of the pipe material, calculate the actual target value, and compare the predicted target value with the actual target value to judge whether the adjustment is effective and reasonable, and perform real-time feedback; wherein the collection frequency of the feedback data is consistent with the parameter acquisition frequency to ensure timely acquisition of the effect information after adjustment;

[0065] The communication module is used to build a high-speed and stable internal communication network through industrial Ethernet technology to realize real-time transmission of data between modules, ensure the timeliness and accuracy of data transmission, control the transmission delay within milliseconds, and ensure the efficiency of the overall operation of the system.

[0066] Further, the identification module includes an anomaly detection unit and a coupling matching unit;

[0067] The abnormality detection unit is used for parameter deviation analysis according to a set analysis interval, combines a moving average method and a trend analysis method, dynamically judges whether each parameter deviates from a normal operation track during operation of the nodular cast iron pipe centrifuge, and introduces a hierarchical judgment mechanism to classify and summarize the deviated parameters, so that different processing measures can be taken according to the severity of the deviation, and the accuracy and reliability of the analysis are ensured. The analysis interval is , which means that parameter deviation analysis is performed once every time period;

[0068] The coupling matching unit is used for analyzing the influence of the deviated parameter on other parameters, obtaining the specific influence correlation parameter, determining which parameters will change due to the change of the deviated parameter, and calculating the influence degree of the deviated parameter on the correlation parameter. In order to obtain the specific influence parameter caused by the deviated parameter, a parameter correlation matrix is established, the parameters affected by the deviated parameter are preliminarily found out through the parameter correlation matrix, the quantitative relationship between the parameters is described based on the parameter correlation matrix and historical data and process knowledge, and a coupling relationship model is established to quantitatively describe the interaction between the parameters.

[0069] Further, the specific steps of the parameter deviation analysis include:

[0070] The time window is set to , a certain specific parameter is set to , the actual values of the parameters in the time window adjacent to the current time point are read from the parameter database, and the moving average value and the standard deviation of the parameter are calculated as the basis for judging the deviation, and the expression is as follows:

[0071]

[0072]

[0073] In the formula, is the moving average value of the current time point , which reflects the average level of the parameter within a certain time, eliminates the influence of short-term fluctuations, and makes the parameter change trend more obvious, is the actual value of the parameter at the time point, is the number of data points in the time window, the time window is the past time points, and , is the downward value, is the sampling interval, and , is the current time point The standard deviation of the moving average value, which measures the dispersion of data relative to the moving average value, reflects the fluctuation range of the parameter;

[0074] The actual value of the current collected parameter is compared with the moving average value , and a standard deviation multiple is set to determine whether the parameter is slightly deviated; in this embodiment, the standard deviation multiple is taken as ;

[0075] If , it indicates that at the current time point , the actual value of the parameter is within the normal fluctuation range centered on the moving average value with a standard deviation multiple as the radius, and it is determined that the parameter is not slightly deviated;

[0076] If , it indicates that at the current time point , the actual value of the parameter is beyond the normal fluctuation range, and it is determined that the parameter is slightly deviated;

[0077] The trend slope of the parameter in the time window is calculated , the actual values of the parameter in a stable production period are selected from the parameter database, and data points are selected, and the historical trend slope is calculated. The trend slope reflects the change rate of the parameter, and comparing the historical trend slope with the current trend slope helps to determine whether the change trend of the current parameter is consistent with the historical normal situation. The expression is as follows:

[0078]

[0079]

[0080] In the formula, the trend slope directly reflects the speed of change of the parameter with time. If , it indicates that the parameter has an upward trend, and if , it indicates that the parameter has a downward trend.

[0081] The deviation threshold is set as , the difference threshold is set as , and the trend difference value of the parameter is calculated ​​, the trend difference reflects the difference between the current trend slope and the historical trend slope to determine whether the parameter exists trend deviation; the expression is as follows:

[0082]

[0083] If and , it indicates that the change trend of the parameter is greatly different from the historical normal situation, and the change amplitude of the current trend exceeds the deviation threshold, an abnormal change trend appears, and it is determined that the parameter exists trend deviation;

[0084] If or , it indicates that the change trend of the parameter is consistent with the historical situation, or although the trend has a certain change, the change amplitude is within an acceptable range, and it is determined that the parameter does not exist trend deviation.

[0085] Further, after the parameter deviation analysis, hierarchical processing measures are implemented according to the severity of the deviation, and the parameters are classified and summarized, which is helpful to targeted processing according to the different states of the parameters, so as to improve the efficiency and accuracy of production management; the specific rules of classification and summary are as follows:

[0086] If the parameter exists both mild deviation and trend deviation, the parameter is defined as a deviation parameter, indicating that the running state of the nodular cast iron pipe centrifuge appears obvious abnormality, for example, the molten iron temperature not only exceeds the normal fluctuation range (mild deviation), but also presents a change trend different from the historical normal trend (trend deviation), which is easy to have adverse effects on the production quality of the nodular cast iron pipe, and needs to be intervened and adjusted in time, and the deviation parameter is transmitted to the coupling matching unit;

[0087] If the parameter does not exist mild deviation and trend deviation, the parameter is defined as a normal parameter, indicating that in the current monitoring period, the parameter is within the normal fluctuation range, the running state of the nodular cast iron pipe centrifuge in the parameter related aspects is stable, and the parameter deviation analysis is continuously performed based on the analysis interval, potential abnormal changes are found in time through continuous monitoring, and it is ensured that the production process is always in a stable state, so as to avoid production accidents or product quality problems caused by sudden parameter abnormalities;

[0088] If the parameter only exists mild deviation or trend deviation, the parameter potential parameter, indicating a parameter The severity of the deviation parameter has not been reached, but there is still a potential risk, which is easy to develop into a deviation parameter in the subsequent production process, with the passage of time or changes in production conditions, and then affect the production quality, for example, the molten iron temperature appears a slight deviation, but the trend is normal, or the trend is abnormal but has not exceeded the normal fluctuation range, at this time the molten iron temperature parameter belongs to the potential parameter, and the potential analysis is continued.

[0089] Further, the specific steps of potential analysis include:

[0090] Based on the sampling interval , the parameter deviation analysis of the potential parameter is continued, and the parameter deviation analysis is performed every time interval to discover the change of the potential parameter in time and prevent the occurrence of production problems;

[0091] After each parameter deviation analysis, the potential parameter is evaluated according to the judgment method of slight deviation and trend deviation; if the analysis result is that there is both slight deviation and trend deviation, the potential parameter is updated to the deviation parameter, otherwise the number of times of deviation (existence of slight deviation or trend deviation) is recorded ;

[0092] After times of parameter deviation analysis, the frequency of deviation of the potential parameter is calculated , and a frequency threshold is set to judge whether the potential parameter is abnormal; the expression is as follows:

[0093]

[0094] If , the frequency of deviation of the potential parameter is high, and there is a high possibility of abnormality, the potential parameter is updated to the deviation parameter;

[0095] If , it is further judged whether there is no deviation for times in succession before the latest parameter deviation analysis; if there is no deviation for times in succession, it indicates that the state of the potential parameter is gradually stable and has returned to the normal level, and the potential parameter is updated to the normal parameter, otherwise it is maintained as the potential parameter and the parameter deviation analysis is continued according to the set time interval.

[0096] Further, the specific steps of calculating the influence degree include:

[0097] The running process of the nodular cast iron pipe centrifuge involves multiple parameters, which are mutually influenced and related. Once a parameter deviates from the normal range, it can easily trigger a series of chain reactions, affecting other related parameters, and then causing adverse effects on the entire production process. Deep mining and analysis of historical production data in the parameter database to determine the correlation between different parameters can help quickly locate other parameters affected when a parameter deviates, so that timely measures can be taken to avoid the expansion of production problems. For example, changes in the temperature of molten iron can affect the fluidity of the molten iron, and then affect the filling time of the pipe mold and the uniformity of the pipe wall thickness. The Pearson correlation coefficient is used to calculate the linear correlation between any two parameters, as shown in the following expression:

[0098]

[0099] wherein, represents the Pearson correlation coefficient between parameters and , , , is the number of parameters, and are the values of parameters and in the first sample, and are the sample means of parameters and , is the sample size; the value range of the correlation coefficient is , the closer the absolute value is to 1, the stronger the linear correlation between the two parameters; the closer the absolute value is to 0, the weaker the correlation;

[0100] According to the correlation between the calculated parameters, a parameter correlation matrix is constructed, wherein the parameter correlation matrix is a square matrix of , which presents the correlation between parameters in a matrix form, provides a clear data structure for subsequent analysis, and through the parameter correlation matrix, the correlation between any two parameters can be quickly queried and judged, providing a convenient tool for analyzing the impact of deviating parameters;

[0101] ​The parameter correlation matrix only reflects the correlation strength and cannot reveal the dynamic change relationship between parameters. In the actual production process, the interaction between parameters is often complex and nonlinear. The change of one parameter is likely to cause chain changes in multiple parameters. Using multiple linear regression analysis, each parameter is taken as the dependent variable and other parameters as independent variables. The regression coefficient is solved by the least squares method to minimize the sum of squares of the error between the predicted value and the actual value, and the regression equation of each parameter is obtained. For example, the tube mold filling time As the independent variable, the dependent variable is the molten iron temperature , centrifuge speed , mold temperature , cooling water flow , generate a preliminary model containing regression equations of various parameters; use process knowledge and expert experience to evaluate and modify the preliminary model, adjust the regression coefficients in each regression equation, make the preliminary model more consistent with the physical and chemical laws of the actual production process, and generate a coupling relationship model ;

[0102] The parameter correlation matrix is ​​used to quickly screen out the initially affected parameter range and set the correlation strength threshold as , when the parameter When it is a deviation parameter, find the parameter in the parameter association matrix The corresponding row ( Rows) and columns ( Column), since the matrix is ​​symmetrical, we only need to consider one row or one column. Taking one row as an example, in the parameter In the corresponding row, check the matrix elements one by one The value of will satisfy Parameters Defined as an associated parameter and saved to the pre-built list of influencing parameters; ,and ;

[0103] The coupling relationship model is used to screen the associated parameters in the influencing parameter list to accurately determine the parameters actually affected. The independent variables of the regression equation with the deviation parameter as the dependent variable in the coupling relationship model are read. For each associated parameter in the influencing parameter list, it is determined whether it is included in these independent variables. If not, the associated parameter is deleted from the influencing parameter list and the influencing parameter list is updated.

[0104] will deviate from the parameters The amount of change Substitute into the coupling relationship model, and calculate the associated parameters in turn for each associated parameter in the influencing parameter list The corresponding change , in order to quantify the impact of deviation parameters on other parameters and provide specific numerical basis for parameter adjustment in the production process; , For parameters The actual value of the current state, For parameters The standard value of is the number of associated parameters, is the change in the associated parameter.

[0105] Furthermore, the specific steps for calculating the optimal operating value include:

[0106] In order to find the optimal operating value under the condition of parameter deviation, an objective function is constructed to quantify and integrate the key indicators in the production process; the product defect rate is directly related to product quality and production efficiency. With minimizing the product defect rate as the optimization goal, the deviation parameters and the associated parameters in the list of influencing parameters are defined as the parameters to be adjusted. According to the basic principles of the production process and the long-term accumulated production experience, the relationship between the product defect rate and the parameters to be adjusted is determined, and the objective function is established. ;

[0107] When calculating the optimal operating value, we cannot arbitrarily select values ​​without considering the actual production situation. There are various constraints in the actual production process. Consider the limitations of production processes and equipment and determine the value range of the parameters to be adjusted. Each parameter to be adjusted has a feasible upper limit in actual production. and lower limit , exceeding the value range will cause production to be unable to proceed normally, or even damage the equipment; based on the coupling relationship model, determine the association constraints between the parameters to be adjusted , and the change It is defined as the adjustable amount of the parameter to be adjusted;

[0108] After the objective function and constraints are given, a set of initial parameter values ​​within the range specified by the constraints is randomly selected to calculate the objective function. Regarding the gradient of each parameter to be adjusted, the value of the parameter to be adjusted is updated according to the iterative formula of gradient descent. In each iteration, check whether the updated value of the parameter to be adjusted meets the constraint conditions. If not, make corresponding adjustments to bring it back to the feasible range. Continue iterating until the value of the objective function converges to a stable minimum value. , and output the corresponding parameter value to be adjusted , and is defined as the best operating value, Defined as the estimated target value.

[0109] Furthermore, the specific steps for real-time feedback include:

[0110] Obtaining total yield of nodular cast iron pipe after parameter adjustment and product qualified amount and calculating actual target value Taking product defect rate as actual target value; expression is as follows:

[0111]

[0112] Setting target difference threshold value as and calculating predicted difference between actual target value and predicted target value to determine whether parameter adjustment is effective; expression is as follows:

[0113]

[0114] If , it indicates that actual effect after current parameter adjustment reaches expectation, and operation continues according to current adjustment strategy, and parameter change is continuously monitored and adjusted in time;

[0115] If , actual effect after current parameter adjustment does not reach expectation, coupling relationship between parameters is re-evaluated, coupling relationship model is updated, and parameter to be adjusted as well as adjustment direction and amplitude are re-determined according to newly evaluated coupling relationship, until actual effect reaches expectation.

[0116] Embodiment 2

[0117] Referring to Figure 5 , another embodiment provided by the application is as follows: a nodular cast iron pipe centrifuge parameter self-adjusting method based on dynamic feedback, comprising the following steps:

[0118] Through various high-precision sensors installed at key positions of the nodular cast iron pipe centrifuge, actual values of various parameters affecting quality of the nodular cast iron pipe in a running process are collected in real time, and the actual values of the parameters are preprocessed, and the preprocessed actual values of the parameters are saved to a constructed parameter database;

[0119] According to a set analysis interval, parameter deviation analysis is performed, dynamic judgment is made on whether various parameters in a running process of the nodular cast iron pipe centrifuge deviate from a normal running track by combining a moving average method and a trend analysis method, a hierarchical judgment mechanism is introduced, deviation parameters are classified and summarized, different processing measures are taken according to severity of deviation, and accuracy and reliability of analysis are ensured;

[0120] Influence of the deviation parameters on other parameters is analyzed, specific associated parameters are obtained, it is determined which parameters will change due to change of the deviation parameters, and influence degree of the deviation parameters on the associated parameters is calculated;

[0121] By reasonably setting the objective function and constraint conditions, the optimal running value and the expected target value of each parameter under the current production conditions are calculated to get rid of the adverse effects caused by parameter deviation and restore the optimal running state, so as to realize the optimization adjustment of the system;

[0122] According to the calculated optimal running value, the parameters of the nodular cast iron pipe centrifuge are adjusted in real time, and after adjusting the parameters, feedback data is continuously collected, including the real-time value of each parameter after adjustment, the quality detection data of the pipe material, the actual target value is calculated, and the expected target value is compared with the actual target value to judge whether the adjustment is effective and reasonable, and real-time feedback is performed.

[0123] Further, the specific steps of parameter deviation analysis include:

[0124] The time window is set to , a certain specific parameter is set to , the actual value of the parameter in the time window adjacent to the current time point is read from the parameter database, and the moving average value and the standard deviation of the parameter are calculated as the basis for judging the deviation, and the expression is as follows:

[0125]

[0126]

[0127] In the formula, is the moving average value of the current time point , reflecting the average level of the parameter within a certain time, eliminating the influence of short-term fluctuations, and making the trend of parameter change more obvious, is the actual value of the parameter at the th time point, is the number of data points in the time window, the time window is the past time points, and , is the lower value, is the sampling interval, and , is the standard deviation of the current time point , which measures the dispersion of the data relative to the moving average value, reflecting the fluctuation amplitude of the parameter;

[0128] The actual value of the currently collected parameter is compared with the moving average value , and a standard deviation multiple is set to judge the parameter whether there is a slight deviation; in this embodiment, take ;

[0129] If , it indicates that at the current time point , the actual value of the parameter is within the normal fluctuation range centered on the moving average value with a standard deviation of times as the radius, it is determined that the parameter does not have a slight deviation;

[0130] If , it indicates that at the current time point , the actual value of the parameter is outside the normal fluctuation range, it is determined that the parameter has a slight deviation;

[0131] Calculate the trend slope of the parameter within the time window , select the actual value of the parameter in a stable production period from the parameter database, assume that data of time points are selected, and calculate the historical trend slope , the trend slope reflects the change rate of the parameter, and comparing the historical trend slope with the current trend slope helps to determine whether the change trend of the current parameter is consistent with the historical normal situation, and the expression is as follows:

[0132]

[0133]

[0134] In the formula, the trend slope directly reflects the speed of change of the parameter with time, if , it indicates that the parameter has an upward trend, and if , it indicates that the parameter has a downward trend;

[0135] Set the deviation threshold to , the difference threshold to , and calculate the trend difference of the parameter , which reflects the difference degree between the current trend slope and the historical trend slope, to determine whether the parameter has a trend deviation; the expression is as follows:

[0136]

[0137] If and , it indicates that the parameter the change trend of the parameter exists trend deviation;

[0138] If or , it indicates that the change trend of the parameter is consistent with the historical situation, or although the trend has a certain change, but the change amplitude is within the acceptable range, then it is judged that the parameter does not exist trend deviation.

[0139] Further, after the parameter deviation analysis, hierarchical processing measures are implemented according to the severity of the deviation, and the parameters are classified and summarized, which is helpful to targeted processing according to the different states of the parameters, so as to improve the efficiency and accuracy of production management; the specific rules of classification and summary are as follows:

[0140] If the parameter exists both mild deviation and trend deviation, the parameter is defined as a deviation parameter, indicating that the running state of the nodular cast iron pipe centrifuge has obvious abnormalities, for example, the molten iron temperature not only exceeds the normal fluctuation range (mild deviation), but also presents a change trend different from the historical normal trend (trend deviation), which is easy to have adverse effects on the production quality of the nodular cast iron pipe, and needs to be intervened and adjusted in time, and the deviation parameter is transmitted to the coupling matching unit;

[0141] If the parameter does not exist mild deviation and trend deviation, the parameter is defined as a normal parameter, indicating that in the current monitoring period, the number is within the normal fluctuation range, the running state of the nodular cast iron pipe centrifuge in the parameter related aspects is stable, and the parameter deviation analysis is continuously performed based on the analysis interval, potential abnormal changes are found in time through continuous monitoring, and it is ensured that the production process is always in a stable state, avoiding production accidents or product quality problems caused by sudden parameter abnormalities;

[0142] If the parameter only exists mild deviation or trend deviation, the parameter is defined as a potential parameter, indicating that the parameter has not reached the severity of the deviation parameter, but there is still a potential risk, which is easy to develop into a deviation parameter in the subsequent production process with the passage of time or the change of production conditions, and then affect the production quality, for example, the molten iron temperature appears mild deviation, but the trend is normal, or the trend is abnormal but has not exceeded the normal fluctuation range, at this time, the molten iron temperature parameter belongs to the potential parameter, and the potential analysis is continuously performed.

[0143] Further, the specific steps of the potential analysis include:

[0144] Based on the sampling interval , the parameter deviation analysis of the potential parameter is continuously performed every time interval to find the change of the potential parameter in time and prevent the occurrence of production problems;

[0145] After each parameter deviation analysis, the potential parameter is evaluated according to the judgment method of the slight deviation and the trend deviation; if the analysis result is that the slight deviation and the trend deviation exist at the same time, the potential parameter is updated as the deviation parameter, otherwise the number of times of the deviation (the slight deviation or the trend deviation) is recorded ;

[0146] After the times of parameter deviation analysis, the frequency of the deviation of the potential parameter is calculated , and a frequency threshold is set to determine whether the potential parameter is abnormal; the expression is as follows:

[0147]

[0148] If , the frequency of the deviation of the potential parameter is high, and there is a high possibility of abnormality, the potential parameter is updated as the deviation parameter;

[0149] If , it is further judged whether the deviation does not occur for times in succession before the latest parameter deviation analysis; if the deviation does not occur for times in succession, it indicates that the state of the potential parameter gradually stabilizes and returns to the normal level, and the potential parameter is updated as the normal parameter, otherwise, the potential parameter is maintained and the parameter deviation analysis is continuously performed according to the set time interval.

[0150] In summary, in the embodiment, the parameters affecting the quality of the pipe are collected in real time, the parameter deviation analysis is performed by using the moving average method and the trend analysis method, the parameters are classified into three categories of deviation, normal and potential according to the deviation degree, the potential parameter is continuously analyzed to prevent its deterioration, the parameter correlation matrix is constructed by using the Pearson correlation coefficient, the coupling relationship model is established by using the multiple linear regression, and the influence degree of the deviation parameter on the related parameter is quantified; the objective function is constructed by minimizing the product defect rate, the production process and equipment constraints are considered, the best running value and the predicted target value are calculated by the gradient descent method, the centrifuge parameters are adjusted in real time according to the values, after the adjustment, the product qualification rate is taken as the actual target value, compared with the predicted target value, the adjustment effect is judged, if the effect is not expected, the coupling relationship is re-evaluated, and the adjustment strategy is optimized.

[0151] The above merely describes the preferred embodiments of the present application, and the protection scope of the present application is not limited to the above-described embodiments. Any technical solution falling within the concept of the present application shall fall within the protection scope of the present application. It should be noted that, for ordinary skilled persons in the art, some improvements and refinements without departing from the principles of the present application shall also be considered as falling within the protection scope of the present application.

Claims

1. A dynamic feedback based centrifuge parameter self-adjusting system for nodular cast iron pipe, characterized in that, The system comprises a data acquisition module, an identification module, an adjustment module and an execution feedback module. The data acquisition module is used for collecting actual values of various parameters affecting the quality of ductile cast iron pipes in real time, pre-processing the actual values of the parameters and saving the actual values into a constructed parameter database. The identification module is used for performing parameter deviation analysis on each parameter in the parameter database, identifying deviation parameters of the ductile cast iron pipe centrifuge, classifying and summarizing the deviation parameters and identifying associated parameters affected by the deviation parameters and the influence degree. The adjustment module is used for calculating optimal operating values and predicted target values of each parameter through a set target function and constraint condition. The execution feedback module is used for adjusting according to the optimal operating values, continuously collecting feedback data, calculating actual target values and performing real-time feedback. The identification module comprises an anomaly detection unit and a coupling matching unit. The anomaly detection unit is used for performing parameter deviation analysis according to a set analysis interval, dynamically determining whether each parameter deviates from a normal operating track during operation of the ductile cast iron pipe centrifuge and classifying and summarizing the parameters. The coupling matching unit is used for analyzing the influence of the deviation parameters on other parameters, obtaining associated parameters and calculating the influence degree. The specific steps of the parameter deviation analysis comprise:

2. The ductile cast iron pipe centrifuge parameter self-adjusting system based on dynamic feedback according to claim 1, characterized in that: Setting a time window as , calculating parameters The moving average in the preset time window and standard deviation , and setting the standard deviation multiple as , determining whether the parameters exist a slight deviation; If then it is determined that the parameter is not deviating slightly; if then it is determined that the parameter is deviating slightly; wherein is the actual value of the parameter at the first time point; computing the parameter a trend slope within the time window and a historical trend slope ; The deviation threshold is set as The difference threshold is The trend difference of the parameter is calculated, and whether the parameter exists trend deviation is judged; the expression is as follows:​ ； If and , it is determined that the parameter has a trend deviation; if or , it is determined that the parameter has no trend deviation. After the parameter deviation analysis, each parameter is classified and summarized, and the specific rules of the classification and summary are as follows: parameters with both slight deviation and trend deviation are defined as deviation parameters, and the deviation parameters are transmitted to the coupling matching unit; parameters without slight deviation and trend deviation are defined as normal parameters, and the parameter deviation analysis is continuously performed; parameters with only slight deviation or trend deviation are defined as potential parameters, and potential analysis is continuously performed. The specific steps of the potential analysis comprise:

3. The dynamic feedback based centrifuge parameter self-adjusting system for nodular cast iron pipe of claim 2, wherein, The specific steps of the calculation of the influence degree comprise: Based on sampling interval continuously performing parameter deviation analysis on the potential parameters; after each parameter deviation analysis, if both a slight deviation and a trend deviation exist, updating the potential parameter to the deviation parameter, otherwise recording the number of deviations ; After performing a secondary parameter deviation analysis, the frequency of deviations is calculated and a frequency threshold is set to determine if the underlying parameter is abnormal; If updating the latent parameter to the deviated parameter; If , before the latest parameter deviation analysis, if the deviation does not occur for consecutive times, the potential parameter is updated as the normal parameter, otherwise, it is kept as the potential parameter, and the parameter deviation analysis according to the sampling interval is continued.

4. The dynamic feedback based centrifuge parameter self-adjusting system for nodular cast iron pipe of claim 3, wherein, each parameter is taken as a dependent variable and other parameters are taken as independent variables to construct a preliminary model, and the preliminary model is evaluated and corrected by using process knowledge and expert experience to generate a coupling relationship model; Based on the parameter database, the Pearson correlation coefficient between any two parameters is calculated, and a parameter correlation matrix is constructed ; The specific steps of the calculation of the optimal operating values comprise: The correlation strength threshold is set as In the parameter correlation matrix, the deviation parameter is traversed The matrix element of the corresponding row will satisfy The parameter is defined as a correlation parameter and saved in the influence parameter list, and the correlation parameter is screened by using the coupling relationship model, and the influence parameter list is updated; wherein, is the deviation parameter The Pearson correlation coefficient between the parameter , , is the number of parameters, and ; The deviation parameter The amount of change Substitute into the coupling relationship model and calculate the corresponding changes of the associated parameters in the influencing parameter list .

5. The dynamic feedback based centrifuge parameter self-adjusting system for nodular cast iron pipe according to claim 4, characterized in that, constraint conditions are set, including the value range of the to-be-adjusted parameters, the adjustable amount and the associated constraint between the to-be-adjusted parameters; The deviation parameter and the associated parameter in the influence parameter list are defined as the parameters to be adjusted, and a target function is established ; The specific steps of the real-time feedback comprise: The parameters to be adjusted are iteratively updated using the gradient descent method until the objective function converges to a stable value. , and output the value of the parameter to be adjusted , and is defined as the best operating value, Defined as the estimated target value.

6. The dynamic feedback based centrifuge parameter self-adjusting system for nodular cast iron pipe of claim 5, wherein, The system comprises: Obtaining total production of nodular cast iron pipes after adjusting parameters and the quantity of products that are qualified and calculating actual target values ; Setting a target difference threshold value as and calculating the predicted difference between the actual target value and the predicted target value ; If , the parameter deviation analysis is continued; If , the actual effect does not reach the expectation, the coupling relationship model is updated, the to-be-adjusted parameter and the optimal operation value are re-determined according to the newly evaluated coupling relationship, and the actual effect reaches the expectation until the actual effect reaches the expectation.

7. A method for dynamic feedback based self-adjustment of parameters of a nodular cast iron pipe centrifuge, implemented based on a dynamic feedback based self-adjustment system of parameters of a nodular cast iron pipe centrifuge as claimed in any one of claims 1 to 6, characterized in that, collecting parameter actual values in real time and saving the actual values into a constructed parameter database; performing parameter deviation analysis according to a set analysis interval and classifying and summarizing the parameters; analyzing the influence of deviation parameters on other parameters, obtaining associated parameters and calculating the influence degree; calculating optimal operating values and predicted target values of each parameter through a set target function and constraint condition; adjusting according to the optimal operating values, continuously collecting feedback data, calculating actual target values and performing real-time feedback. ​

Images