A Method and System for Predicting the Mechanical Properties of a Cast Pump Body Based on Processing Parameters

Abstract

The present invention relates to a method and system for predicting the mechanical properties of a casting pump body based on processing parameters; collecting the processing parameters of the current pump body casting, and collecting the composition fluctuation parameters of the casting within a recently preset time range; indexing and obtaining the processing error parameters according to the processing parameters; collecting the fluid characteristic parameters in the pump body operation scenario, and analyzing to obtain the fluid complexity coefficient and the operation quality requirement coefficient; configuring a stepped prediction resource according to the composition fluctuation parameters, the processing error parameters, the fluid complexity coefficient and the operation quality requirement coefficient, and performing prediction on the mechanical properties of the casting pump body, so as to predict and obtain the mechanical property parameters of the pump body. The present invention solves the technical problem in the prior art that the mechanical properties of a casting pump body cannot be scientifically and accurately predicted based on processing parameters.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and particularly relates to a method and system for predicting the mechanical properties of a cast pump body based on processing parameters. Background Art

[0002] In the field of industrial production, the pump body is a key component in various mechanical devices, and its performance directly affects the operating efficiency and stability of the entire device. The casting process is a common method for producing pump bodies. However, the casting process is affected by numerous processing parameters, making it difficult to accurately control the mechanical properties of the pump body. Currently, there is a lack of a comprehensive and accurate method and system for predicting the mechanical properties of cast pump bodies. Often, experienced technicians make intuitive judgments, which are unscientific and inaccurate. Summary of the Invention

[0003] Aiming at the technical problems of unscientific and inaccurate prediction of the mechanical properties of pump bodies in the prior art, the present invention provides a method and system for predicting the mechanical properties of a cast pump body based on processing parameters to solve this problem.

[0004] The technical solution of the present invention to solve the above technical problems is as follows:

[0005] In a first aspect, the present invention provides a method for predicting the mechanical properties of a cast pump body based on processing parameters, including: collecting the processing parameters of the current pump body casting, and collecting the composition fluctuation parameters of the casting in the most recent preset time range; indexing and obtaining the processing error parameters according to the processing parameters; collecting the fluid characteristic parameters in the pump body operation scenario, and analyzing to obtain the fluid complexity coefficient and the operation quality requirement coefficient; configuring stepped prediction resources according to the composition fluctuation parameters, processing error parameters, fluid complexity coefficient and operation quality requirement coefficient, and performing prediction on the mechanical properties of the cast pump body to obtain the mechanical property parameters of the pump body.

[0006] Optionally, collecting the processing parameters of the current pump body casting, and collecting the composition fluctuation parameters of the casting in the most recent preset time range includes: collecting the processing parameters of the current pump body casting, where the processing parameters include the pouring temperature, cooling rate, and mold preheating temperature; collecting the composition parameters of the pump body casting in the most recent preset time range to obtain a historical composition parameter sequence; and processing the historical composition parameter sequence to obtain the composition fluctuation parameters.

[0007] Among them, processing to obtain a component fluctuation parameter according to the historical component parameter sequence includes: calculating the mean of the historical component parameter sequence to obtain an average historical component parameter; randomly selecting several historical component parameters within the historical component parameter sequence, and respectively calculating the absolute error magnitude from the average historical component parameter to obtain a plurality of component absolute error magnitudes; calculating the mean of the plurality of component absolute error magnitudes to obtain a component fluctuation parameter.

[0008] Optionally, indexing to obtain a processing error parameter according to the processing parameter includes: calling a processing parameter library, where the processing parameter library is constructed based on historical processing parameters and includes a sample processing parameter set and a sample processing error parameter set, and each processing error parameter is the average error magnitude when different sample processing parameters are controlled; inputting the processing parameter into the processing parameter library to index and obtain a processing error parameter.

[0009] Optionally, collecting fluid characteristic parameters in the pump body operation scenario and analyzing to obtain a fluid complexity coefficient and an operation quality requirement coefficient includes: collecting fluid parameters in the pump body operation scenario to obtain a fluid parameter sequence; calculating the mean of the fluid parameter sequence to obtain an average fluid parameter; randomly selecting several fluid parameters within the fluid parameter sequence, respectively calculating the absolute error magnitude from the average fluid parameter and calculating the mean to obtain a fluid complexity coefficient; using the average fluid parameter as the operation quality requirement coefficient.

[0010] Optionally, configuring stepped prediction resources according to the component fluctuation parameter, processing error parameter, fluid complexity coefficient, and operation quality requirement coefficient, and performing mechanical property prediction of the cast pump body, and predicting to obtain the mechanical property parameters of the pump body, including: training an array of stepped pump body performance predictors, where the array of stepped pump body performance predictors includes a plurality of pump body performance predictors corresponding to a plurality of performance prediction coefficient intervals, and the training data volume of each pump body performance predictor is different; calculating a performance prediction coefficient according to the component fluctuation parameter, processing error parameter, fluid complexity coefficient, and operation quality requirement coefficient; selecting a corresponding pump body performance predictor according to the performance prediction coefficient interval in which the performance prediction coefficient falls, inputting the processing parameter, and predicting and outputting to obtain the mechanical property parameters of the pump body.

[0011] Among them, training the stepped pump body performance predictor array includes: collecting a set of sample processing parameters and a set of sample mechanical property parameters according to the pump body casting data within the historical time; constructing multiple pump body performance predictors by using machine learning; dividing the set of sample processing parameters and the set of sample mechanical property parameters to obtain multiple portions of pump body performance prediction training data with different data volumes, and the data volumes of the multiple portions of pump body performance prediction training data are distributed from large to small; respectively using the multiple portions of pump body performance prediction training data to perform supervised training and testing on the multiple pump body performance predictors until convergence; obtaining multiple performance prediction coefficient intervals, where the performance prediction coefficient is greater than or equal to 0 and less than or equal to 1; constructing a mapping relationship with the multiple performance prediction coefficient intervals according to the training data volumes of the multiple pump body performance predictors, where the size of the performance prediction coefficient interval is positively correlated with the training data volume of the pump body performance predictor.

[0012] Among them, calculating the performance prediction coefficient according to the composition fluctuation parameter, processing error parameter, fluid complexity coefficient, and operation quality requirement coefficient includes: collecting the maximum composition fluctuation parameter, maximum processing error parameter, maximum fluid complexity coefficient, and maximum operation quality requirement coefficient within the historical time according to the historical data of pump body casting; respectively calculating the ratios of the composition fluctuation parameter, processing error parameter, fluid complexity coefficient, and operation quality requirement coefficient to the maximum composition fluctuation parameter, maximum processing error parameter, maximum fluid complexity coefficient, and maximum operation quality requirement coefficient, and calculating the performance prediction coefficient by weighted calculation.

[0013] In a second aspect, the present invention provides a system for predicting the mechanical properties of a cast pump body based on processing parameters, including:

[0014] A processing parameter acquisition module, configured to acquire the processing parameters for current pump body casting and acquire the composition fluctuation parameters for casting within the most recent preset time range;

[0015] A processing error parameter indexing module, configured to index and obtain the processing error parameter according to the processing parameter;

[0016] A fluid characteristic parameter analysis module, configured to acquire the fluid characteristic parameters in the pump body operation scenario and analyze to obtain the fluid complexity coefficient and the operation quality requirement coefficient;

[0017] A mechanical property prediction module, configured to configure stepped prediction resources according to the composition fluctuation parameter, processing error parameter, fluid complexity coefficient, and operation quality requirement coefficient, perform prediction on the mechanical properties of the cast pump body, and predict to obtain the mechanical property parameters of the pump body.

[0018] In the present invention, by collecting the processing parameters of the current pump body casting and the composition fluctuation parameters of the casting in the most recent preset time range, and evaluating the credibility of the prediction result according to the composition fluctuation parameters, a pump body mechanical property predictor with different training amounts is configured, realizing the precise allocation of computing resources; in the present invention, according to the processing parameters, the processing error parameters are indexed and obtained, and the error factors in the processing process are introduced into the prediction process as weight factors, improving the scientific nature of the prediction result; in the present invention, the fluid characteristic parameters in the pump body operation scenario are collected, and the fluid complexity coefficient and the operation quality requirement coefficient are analyzed, realizing the configuration of a pump body mechanical property predictor with different training amounts according to different usage scenarios of the pump body, increasing the credibility of the prediction result; in the present invention, according to the composition fluctuation parameters, the processing error parameters, the fluid complexity coefficient and the operation quality requirement coefficient, a stepped prediction resource is configured to predict the mechanical properties of the casting pump body, and the mechanical property parameters of the pump body are obtained by prediction. The weights of various parameters are comprehensively calculated for the allocation of computing resources, and the computing resources are precisely configured according to different prediction tasks, so as to achieve the effect of saving computing power and improving the prediction efficiency.

[0019] In summary, by implementing the present application, the technical effect of scientifically and accurately predicting the mechanical properties of the casting pump body according to the processing parameters can be achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 is a schematic flow chart of a method for predicting the mechanical properties of a casting pump body based on processing parameters provided by the present invention;

[0021] Figure 2 is a schematic structural diagram of a system for predicting the mechanical properties of a casting pump body based on processing parameters provided by the present invention.

[0022] In the drawings, the components represented by the reference numerals are as follows:

[0023] The processing parameter acquisition module 11, the processing error parameter indexing module 12, the fluid characteristic parameter analysis module 13, and the mechanical property prediction module 14. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0024] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative efforts shall fall within the protection scope of the present invention.

[0025] In the description of the present invention, the terms "first" and "second" are used for descriptive purposes only and cannot be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include one or more of the said features. In the description of the present invention, "a plurality of" means two or more unless otherwise specifically defined.

[0026] In the description of the present invention, the term "for example" is used to mean "serving as an example, illustration, or explanation". Any embodiment described as "for example" in the present invention is not necessarily construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to implement and use the present invention. In the following description, details are set forth for purposes of explanation. It should be understood that those of ordinary skill in the art can recognize that the present invention can be implemented without these specific details. In other instances, well-known structures and processes are not elaborated in detail to avoid obscuring the description of the present invention with unnecessary details. Therefore, the present invention is not intended to be limited to the embodiments shown, but is consistent with the broadest scope that conforms to the principles and features disclosed in the present invention.

[0027] Embodiment 1, as Figure 1 shown, an embodiment of the present invention provides a method for predicting the mechanical properties of a casting pump body based on processing parameters, including:

[0028] S100: Collect the processing parameters of the current pump body casting and collect the component fluctuation parameters of the casting in the most recent preset time range;

[0029] S200: Index and obtain the processing error parameters according to the said processing parameters;

[0030] S300: Collect the fluid characteristic parameters in the pump body operation scenario, and analyze to obtain the fluid complexity coefficient and the operation quality requirement coefficient;

[0031] S400: Configure stepped prediction resources according to the said component fluctuation parameters, processing error parameters, fluid complexity coefficient, and operation quality requirement coefficient, and perform prediction on the mechanical properties of the casting pump body to obtain the mechanical property parameters of the pump body.

[0032] In step S100 of the embodiment of the present application, collecting the processing parameters of the current pump body casting and collecting the component fluctuation parameters of the casting in the most recent preset time range includes:

[0033] Collect the processing parameters of the current pump body casting, where the said processing parameters include the pouring temperature, cooling rate, and mold preheating temperature;

[0034] Collect the composition parameters of the pump body casting within the recently preset time range to obtain a historical composition parameter sequence;

[0035] Process and obtain the composition fluctuation parameters according to the historical composition parameter sequence.

[0036] During the casting process, the processing parameters of the pump body have a great impact on the performance of the finally cast pump body. In the embodiments of the present application, collecting the processing parameters of the current pump body casting is to obtain the basic situation of the currently cast pump body as the basis for predicting the performance of the cast pump body. The parameters that need to be collected include the pouring temperature, cooling rate, and mold preheating temperature. Among them, the pouring temperature refers to the temperature when the molten metal enters the mold for filling from the pressure chamber. Improper pouring temperature will cause defects such as shrinkage cavities, pores, deformation, and cracks in the casting, thus affecting the performance of the casting; the cooling rate refers to the speed at which the molten metal cools down during the solidification process. The cooling rate will determine parameters such as the grain size, microstructure, and residual stress of the casting, thus affecting the performance of the casting; and the mold temperature generally refers to the surface temperature of the mold. The mold temperature greatly affects the mechanical properties, dimensional accuracy, and service life of the die-casting mold. Among them, the cooling rate can be measured by the amplitude of temperature drop per unit time. For example, a temperature drop of 35 degrees Celsius within 1 minute can be expressed as 35℃ / min.

[0037] In the actual production process, in addition to the above-mentioned processing parameters, the material composition of the casting also has a great impact on the casting performance. The material properties of different compositions vary widely. Even for a material with a dominant content of a certain component (such as a single material accounting for 99% in a certain alloy), the performance of this material may fluctuate violently due to the fluctuation of the amount of a rare earth element with a very small addition amount. Therefore, in the embodiments of the present application, the material composition of the pump body is also a factor that must be considered. Therefore, it is necessary to collect the composition parameters of the pump body casting within the recently preset time range. The preset time can be determined according to the actual output of the factory. For example, if the output is large, sufficient data can be obtained within a relatively short preset time (such as 1 month). If the output is small, it takes a longer time (such as 1 year) to collect sufficient sample data. The method for obtaining the composition parameters of the pump body casting can be to take samples of the pump body after each casting is completed, and then use methods such as atomic emission spectrometry or atomic absorption spectrometry to measure the elemental content ratio (such as iron, carbon, silicon, manganese, etc.) in the pump body material. By the above method, collecting the composition parameters of the pump body casting within the recently preset time range, a historical composition parameter sequence can be obtained.

[0038] Among them, processing and obtaining the composition fluctuation parameters according to the historical composition parameter sequence includes:

[0039] Calculate the mean value of the historical composition parameter sequence to obtain the average historical composition parameter;

[0040] Randomly select several historical component parameters within the historical component parameter sequence, calculate the absolute error magnitudes with respect to the average historical component parameter respectively, and obtain multiple component absolute error magnitudes;

[0041] Calculate the mean of the multiple component absolute error magnitudes to obtain the component fluctuation parameter.

[0042] In the embodiment of the present application, calculating the mean of the historical component parameter sequence is to calculate, based on this, the possible magnitude of the fluctuation of the pump body components during the actual production process. The possible magnitude of the fluctuation is measured by the component fluctuation parameter. Specifically, the mean of the historical component parameter sequence is the mean of the elements in the historical component parameter sequence. For example, there are three sets of component parameters within a preset time, and the iron contents are 99%, 98%, and 97% respectively. Then the mean of the iron element in this component parameter sequence is (99% + 98% + 97%) / 3 = 98%. By this method, the mean of the historical component parameter sequences of all elements can be calculated as the average historical component parameter.

[0043] Further, after obtaining the mean of the historical component parameter sequence, it is necessary to calculate the possible magnitude of the fluctuation of the pump body components based on this. The calculation method is to randomly select several historical component parameters within the historical component parameter sequence, calculate the absolute error magnitudes with respect to the average historical component parameter respectively, and obtain multiple component absolute error magnitudes. For example, the average historical component parameter of the iron element is 98%, and the randomly selected several historical component parameters are 99%, 98%, and 97%. It is easy to see that the absolute error magnitudes of these historical component parameters with respect to the average historical component parameter (taking the absolute value after subtraction) are 1%, 0%, and 1%, and the mean of the multiple component absolute error magnitudes is (1% + 0% + 1%) / 3 ≈ 0.667%. According to the above calculation method, the component fluctuation parameters of the contents of multiple elements in the cast pump body can be calculated, and the mean can be calculated as the final component fluctuation parameter.

[0044] In step S200 of the embodiment of the present application, indexing to obtain the processing error parameter according to the processing parameter includes:

[0045] Call the processing parameter library, where the processing parameter library is constructed based on historical processing parameters and includes a sample processing parameter set and a sample processing error parameter set. Each processing error parameter is the average error magnitude when different sample processing parameters are controlled;

[0046] Input the processing parameter into the processing parameter library and index to obtain the processing error parameter.

[0047] In the actual production process of a casting pump body, due to reasons such as environmental temperature changes, equipment aging, and irregular personnel operations, there will be certain deviations between the preset actual pouring temperature, cooling rate, and mold preheating temperature and the preset parameters. In the embodiments of the present application, the introduction of processing error parameters is to estimate the difference between the actual performance and the predicted performance of the casting pump body produced under the processing parameters based on the error range between the actual value and the preset value of the processing parameters in the previous actual production process, so as to judge the credibility of the prediction result. Specifically, the processing error range under the historical parameters identical to the processing parameters to be predicted can be indexed through a pre-established processing parameter library. The database contains two types of data. One is the sample processing parameters, that is, the preset processing parameters. For example, a pouring temperature of 600 °C, a cooling rate of 30 °C / min, and a mold preheating temperature of 200 °C can form a set of sample processing parameters. The other type of data is the sample processing error parameters, which are the average error ranges when different sample processing parameters are controlled. For example, within a certain period of time, 3 casting pump bodies were produced with the preset processing parameters of a pouring temperature of 600 °C, a cooling rate of 30 °C / h, and a mold preheating temperature of 200 °C. The actual pouring temperatures during the 3 productions were 603 °C, 597 °C, and 600 °C respectively. Then the error ranges are the absolute values of the differences from 600 °C, which are 3 °C, 3 °C, and 0 °C respectively. The average error range is (3 °C + 3 °C + 0 °C) / 3 = 2 °C, and the error range is 0.33%, which is used as the processing error parameter. By using this method, the average processing error parameters of multiple processing parameters can be calculated.

[0048] Among them, since the temperature measurement element of the equipment itself is no longer reliable when measuring the processing error parameters, a specially calibrated temperature detection device needs to be used for measurement when measuring the sample processing error parameters.

[0049] Collect multiple sample processing parameters and the corresponding sample processing error parameters of the sample processing parameters to obtain a sample processing parameter set and a sample processing error parameter set, constituting a processing parameter library. Among them, the processing parameter library should contain all common processing parameters and the corresponding processing error parameters. As long as the processing parameters are input, the corresponding processing error parameters can be output. The construction method of this processing parameter library is prior art and will not be elaborated here.

[0050] In step S300 of the embodiments of the present application, the fluid characteristic parameters in the pump body operation scenario are collected, and the fluid complexity coefficient and the operation quality requirement coefficient are analyzed, including:

[0051] Collect the fluid parameters in the pump body operation scenario to obtain a fluid parameter sequence;

[0052] Calculate the mean value of the fluid parameter sequence to obtain the average fluid parameter;

[0053] Randomly selecting a number of fluid parameters in the fluid parameter sequence, calculating the absolute error margins from the average fluid parameter, and calculating the average to obtain a fluid complexity coefficient;

[0054] The average fluid parameter is used as the operation quality requirement coefficient.

[0055] In the embodiment of the present application, the fluid complexity coefficient and the operation quality requirement coefficient are introduced in order to evaluate the use environment of the pump body under normal operation, to evaluate the degree of its demand for casting quality, and to determine whether the accuracy of the prediction of the physical properties of the pump body can meet the demand. Optionally, the fluid parameters in the pump body operation scenario can be common fluid parameters such as pressure and head. Taking pressure as an example, a pressure sensor can be built into the pump body or the inner wall of the pipe connected to the pump body to collect the fluid pressure that the pump body is subjected to during normal operation. The unit of pressure is usually MPa. The frequency of fluid parameter collection can be determined according to the specific use scenario of the pump body. For example, a civil water pump can set the frequency of fluid parameter collection to once per hour, collect data a total of 24 times, and obtain a fluid parameter sequence. Then it is necessary to calculate the mean of the fluid parameter sequence to obtain the average fluid parameter; then randomly select a number of fluid parameters in the fluid parameter sequence, calculate the absolute error amplitude with the average fluid parameter, and calculate the mean to obtain the fluid complexity coefficient. For example, if the mean of a sequence of pump fluid parameters (such as pressure) collected over a period of time is 1 MPa, and several fluid parameters are randomly selected from the sequence as 1.3 MPa, 1.0 MPa, and 0.7 MPa, then the absolute errors between these three fluid parameters and the average fluid parameter are 0.3 MPa, 0 MPa, and 0.3 MPa, and the mean absolute error between these three fluid parameters and the average fluid parameter is 0.2 MPa. The mean absolute error between these three fluid parameters and the average fluid parameter is the fluid complexity coefficient, i.e., 0.2 MPa / 1 MPa = 20%. The mean value of the sequence of pump fluid parameters (such as pressure) collected over the aforementioned period of time (e.g., 1 MPa) is the operation quality requirement coefficient.

[0056] In step S400 of the embodiment of the present application, based on the composition fluctuation parameter, processing error parameter, fluid complexity coefficient, and operation quality requirement coefficient, a step-by-step prediction resource is configured to predict the mechanical properties of the casting pump body, and the mechanical performance parameters of the pump body are obtained by prediction, including:

[0057] Training a stepped pump performance predictor array, wherein the stepped pump performance predictor array includes a plurality of pump performance predictors corresponding to a plurality of performance prediction coefficient intervals, and each pump performance predictor has a different amount of training data;

[0058] Calculating a performance prediction coefficient based on the composition fluctuation parameter, the processing error parameter, the fluid complexity coefficient, and the operation quality requirement coefficient;

[0059] According to the performance prediction coefficient interval in which the performance prediction coefficient falls, select the corresponding pump body performance predictor, input the processing parameters, and predict and output to obtain the mechanical property parameters of the pump body.

[0060] Among them, training the stepped pump body performance predictor array includes:

[0061] According to the pump body casting data within the historical time, collect the sample processing parameter set and collect the sample mechanical property parameter set;

[0062] Adopt machine learning to construct multiple pump body performance predictors;

[0063] Divide the sample processing parameter set and the sample mechanical property parameter set to obtain multiple pump body performance prediction training data with different data volumes, and the data volumes of the multiple pump body performance prediction training data are distributed from large to small;

[0064] Respectively use the multiple pump body performance prediction training data to conduct supervised training and testing on the multiple pump body performance predictors until convergence;

[0065] Obtain multiple performance prediction coefficient intervals, where the performance prediction coefficient is greater than or equal to 0 and less than or equal to 1;

[0066] Construct the mapping relationship with multiple performance prediction coefficient intervals according to the training data volumes of the multiple pump body performance predictors, where the size of the performance prediction coefficient interval is positively correlated with the training data volume of the pump body performance predictor.

[0067] In the embodiments of the present application, the sample processing parameter set is the pouring temperature, cooling rate, and mold preheating temperature of the cast pump body described in S100, and the sample mechanical property parameters are a series of parameters that can reflect the quality of the pump body, including yield strength (MPa), Vickers hardness (HV), etc. By collecting the processing parameters of multiple cast pump bodies and the mechanical property parameters of the corresponding finished cast pump bodies under these parameters, the sample processing parameter set and the sample mechanical property parameter set can be obtained. Divide the sample processing parameter set and the sample mechanical property parameter set into a training set and a validation set according to a ratio of 8:2 for training the stepped pump body performance predictor array.

[0068] The input of this pump body performance predictor is the processing parameters (pouring temperature, cooling rate, mold preheating temperature), and the output is the predicted mechanical properties (yield strength, Vickers hardness) under this processing parameter. Therefore, when building the pump body performance predictor, a deep neural network (DNN) can be used for building. The DNN can effectively capture the non-linear relationship between the casting process parameters and the mechanical properties and perform stably in multi-output prediction tasks.

[0069] Optionally, the input layer of the pump body performance predictor has 3 nodes (pouring temperature, cooling rate, mold preheating temperature). The hidden layer is divided into three layers. The first layer has 128 nodes and the activation function is ReLU (to alleviate gradient vanishing); the second layer has 64 nodes and the activation function is ReLU; the third layer has 32 nodes and the activation function is Swish (to enhance non-linear expression ability). The output layer has 2 nodes (yield strength, Vickers hardness) and the activation function is linear. In terms of model hyperparameter settings, the Adam optimizer is used (learning rate = 0.001, β1 = 0.9, β2 = 0.999); the mean squared error (MSE) is selected as the loss function; L2 regularization is selected for regularization (coefficient = 0.001) to prevent overfitting; the batch size for each training is set to 32 (to balance training speed and gradient stability); the number of training epochs is initially set to 200 epochs, and an early stopping mechanism is added (patience = 10, that is, if the validation loss does not decrease for 10 consecutive epochs, the training is terminated). When the decrease in the validation set loss (MSE) is < 0.1% for 5 consecutive epochs, it is determined that the model converges, and the pump body performance predictor is obtained.

[0070] In the embodiments of the present application, by controlling the size of the training sample size, through the above-mentioned pump body performance predictor training method, multiple pump body performance predictors can be obtained. For example, the aforementioned sample processing parameter set and sample mechanical property parameter set can be divided into three parts, each containing 3000 sets of data, 2000 sets of data, and 1000 sets of data, and the first pump body performance predictor, the second pump body performance predictor, and the third pump body performance predictor are respectively obtained through training. They are used to handle pump body performance prediction tasks under different performance prediction coefficient intervals. For example, the first pump body performance predictor is used to handle performance prediction tasks with a performance prediction coefficient interval of 0 to 0.4, the second pump body performance predictor is used to handle performance prediction tasks with a performance prediction coefficient interval of 0.4 to 0.7, and the third pump body performance predictor is used to handle performance prediction tasks with a performance prediction coefficient interval of 0.7 to 1. Since the performance prediction coefficient represents the difficulty and importance of performance prediction, the larger the performance prediction coefficient interval, the larger the amount of training data allocated to the pump body performance predictor. In this way, it is possible to accurately allocate computing resources according to different prediction tasks, so as to save computing power and improve prediction efficiency and accuracy.

[0071] Among them, calculating the performance prediction coefficient according to the composition fluctuation parameter, processing error parameter, fluid complexity coefficient, and operation quality requirement coefficient includes:

[0072] According to the historical data of pump body casting, collect the maximum composition fluctuation parameter, maximum processing error parameter, maximum fluid complexity coefficient, and maximum operation quality requirement coefficient within the historical time;

[0073] Calculate the ratios of the component fluctuation parameter, processing error parameter, fluid complexity coefficient, and operation quality requirement coefficient to the maximum component fluctuation parameter, maximum processing error parameter, maximum fluid complexity coefficient, and maximum operation quality requirement coefficient respectively, and calculate the performance prediction coefficient through weighted calculation;

[0074] In the embodiment of the present application, the performance prediction coefficient is an index reflecting the difficulty and importance of performance prediction. Through the magnitude of the performance prediction coefficient, the pump body performance predictors trained with different amounts of training data can be reasonably configured for pump body performance prediction.

[0075] Through the method in the foregoing steps, the average component fluctuation parameter (such as 2%) in the historical time, the average processing error parameter (such as the pouring temperature error parameter of 2°C, the cooling rate error parameter of 3°C / h, and the mold preheating temperature error parameter of 3°C), the average fluid complexity coefficient (such as 0.2MP), and the average operation quality requirement coefficient (such as 1MPa) can be obtained;

[0076] Calculate the ratios of the above-mentioned component fluctuation parameter, processing error parameter, fluid complexity coefficient, and operation quality requirement coefficient to the maximum component fluctuation parameter (such as 4%), maximum processing error parameter (such as the pouring temperature error parameter of 4°C, the cooling rate error parameter of 6°C / h, and the mold preheating temperature error parameter of 6°C), maximum fluid complexity coefficient (such as 0.4MP), and maximum operation quality requirement coefficient (such as 2MPa) respectively, and the component fluctuation parameter ratio can be obtained as 2% / 4% = 0.5, the pouring temperature error parameter ratio as 2°C / 4°C = 0.5, the cooling rate error parameter ratio as (3°C / h) / (6°C / h) = 0.5, the mold preheating temperature error parameter ratio as 3°C / 6°C = 0.5, the fluid complexity coefficient ratio as 0.2MP / 0.4MP = 0.5, and the operation quality requirement coefficient ratio as 1MPa / 2MPa = 0.5. Take the average of all the above ratios to obtain the performance prediction coefficient as (0.5 + 0.5 + 0.5 + 0.5 + 0.5 + 0.5) / 6 = 0.5. The historical time can be determined as appropriate according to the specific production situation (such as 6 months), as long as the data obtained within the historical time is sufficiently representative.

[0077] Finally, according to the performance prediction coefficient interval into which the performance prediction coefficient falls, select the corresponding pump body performance predictor, input the processing parameters (pouring temperature, cooling rate, mold preheating temperature), and the mechanical property parameters (such as yield strength, Vickers hardness) of the pump body can be predicted and output.

[0078] Embodiment 2, as Figure 2 shown, based on the same inventive concept as the method for predicting the mechanical properties of a cast pump body based on processing parameters provided in Embodiment 1, the embodiment of the present invention further provides a system for predicting the mechanical properties of a cast pump body based on processing parameters, including:

[0079] The processing parameter acquisition module 11 is used to acquire the processing parameters for current pump body casting and the composition fluctuation parameters for casting within a recently preset time range;

[0080] The processing error parameter indexing module 12 is used to index and obtain processing error parameters according to the processing parameters;

[0081] The fluid characteristic parameter analysis module 13 is used to acquire the fluid characteristic parameters in the pump body operation scenario and analyze to obtain the fluid complexity coefficient and the operation quality requirement coefficient;

[0082] The mechanical property prediction module 14 is used to configure stepped prediction resources according to the composition fluctuation parameters, processing error parameters, fluid complexity coefficient, and operation quality requirement coefficient, perform mechanical property prediction of the cast pump body, and predict to obtain the mechanical property parameters of the pump body.

[0083] Furthermore, the processing parameter acquisition module 11 is also used for:

[0084] Acquire the processing parameters for current pump body casting, where the processing parameters include the pouring temperature, cooling rate, and mold preheating temperature;

[0085] Acquire the composition parameters for pump body casting within a recently preset time range to obtain a historical composition parameter sequence;

[0086] Process and obtain the composition fluctuation parameters according to the historical composition parameter sequence.

[0087] Among them, processing and obtaining the composition fluctuation parameters according to the historical composition parameter sequence includes:

[0088] Calculate the mean value of the historical composition parameter sequence to obtain the average historical composition parameter;

[0089] Randomly select several historical composition parameters within the historical composition parameter sequence, and calculate the absolute error magnitudes with respect to the average historical composition parameter respectively to obtain multiple composition absolute error magnitudes;

[0090] Calculate the mean value of the multiple composition absolute error magnitudes to obtain the composition fluctuation parameter.

[0091] Furthermore, the processing error parameter indexing module 12 is also used for:

[0092] Call the processing parameter library, where the processing parameter library is constructed based on historical processing parameters and includes a sample processing parameter set and a sample processing error parameter set, and each processing error parameter is the average error magnitude when different sample processing parameters are controlled;

[0093] Input the processing parameters into the processing parameter library, and obtain the processing error parameters through indexing.

[0094] Furthermore, the fluid characteristic parameter analysis module 13 is also used for:

[0095] Collect the fluid parameters in the pump body operation scenario to obtain a fluid parameter sequence;

[0096] Calculate the mean value of the fluid parameter sequence to obtain the average fluid parameter;

[0097] Randomly select several fluid parameters within the fluid parameter sequence, calculate the absolute error magnitude with the average fluid parameter respectively, and calculate the mean value to obtain the fluid complexity coefficient;

[0098] Take the average fluid parameter as the operation quality requirement coefficient.

[0099] Furthermore, the mechanical property prediction module 14 is also used for:

[0100] Train a stepped pump body performance predictor array, where the stepped pump body performance predictor array includes multiple pump body performance predictors, corresponding to multiple performance prediction coefficient intervals, and the training data volume of each pump body performance predictor is different;

[0101] Calculate the performance prediction coefficient based on the composition fluctuation parameter, processing error parameter, fluid complexity coefficient and operation quality requirement coefficient;

[0102] According to the performance prediction coefficient interval into which the performance prediction coefficient falls, select the corresponding pump body performance predictor, input the processing parameters, and predict and output to obtain the mechanical property parameters of the pump body.

[0103] Among them, training the stepped pump body performance predictor array includes:

[0104] According to the pump body casting data within the historical time, collect the sample processing parameter set and the sample mechanical property parameter set;

[0105] Adopt machine learning to construct multiple pump body performance predictors;

[0106] Divide the sample processing parameter set and the sample mechanical property parameter set to obtain multiple portions of pump body performance prediction training data with different data volumes, and the data volumes of the multiple portions of pump body performance prediction training data are distributed from large to small;

[0107] Respectively use the multiple portions of pump body performance prediction training data to conduct supervised training and testing on the multiple pump body performance predictors until convergence;

[0108] Obtain multiple performance prediction coefficient intervals, where the performance prediction coefficient is greater than or equal to 0 and less than or equal to 1;

[0109] Construct a mapping relationship with multiple performance prediction coefficient intervals according to the training data volume of the multiple pump body performance predictors, where the size of the performance prediction coefficient interval is positively correlated with the training data volume of the pump body performance predictor.

[0110] Among them, the performance prediction coefficient is calculated according to the composition fluctuation parameter, processing error parameter, fluid complexity coefficient, and operation quality requirement coefficient, including:

[0111] According to the historical data of pump body casting, collect the maximum composition fluctuation parameter, maximum processing error parameter, maximum fluid complexity coefficient, and maximum operation quality requirement coefficient within the historical time;

[0112] Calculate the ratios of the composition fluctuation parameter, processing error parameter, fluid complexity coefficient, and operation quality requirement coefficient to the maximum composition fluctuation parameter, maximum processing error parameter, maximum fluid complexity coefficient, and maximum operation quality requirement coefficient respectively, and calculate the performance prediction coefficient by weighted calculation.

[0113] It should be noted that in the above embodiments, the descriptions of each embodiment have their own emphases. For the parts not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0114] Those skilled in the art should understand that the embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0115] The present invention is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, and the combination of processes and / or blocks in the flowcharts and / or block diagrams, can be realized by computer program instructions. These computer program instructions can be provided to the processors of general-purpose computers, special-purpose computers, embedded computers, or other programmable data processing devices to generate a machine, so that the instructions executed by the processors of the computer or other programmable data processing devices generate means for realizing the functions specified in one Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0116] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means that implement the function specified in one or more of the procedures Figure 1 or blocks Figure 1 specified in one or more of the procedures and / or blocks.

[0117] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, whereby the instructions executed on the computer or other programmable apparatus provide steps for implementing the function specified in one or more of the procedures Figure 1 or blocks Figure 1 specified in one or more of the blocks.

[0118] Although the preferred embodiments of the present invention have been described, additional changes and modifications can be made by those skilled in the art once they learn of the basic inventive concept.

[0119] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the present invention and its equivalent technologies, the present invention is also intended to include these modifications and variations.

Claims

1. A method for predicting the mechanical properties of a casting pump body based on processing parameters, characterized in that The method includes: Collecting the processing parameters for the current pump body casting, and collecting the composition fluctuation parameters for the casting within a recent preset time range, including: Collecting the processing parameters for the current pump body casting, where the processing parameters include the pouring temperature, cooling rate, and mold preheating temperature; Collecting the composition parameters for the pump body casting within a recent preset time range to obtain a historical composition parameter sequence; Processing the historical composition parameter sequence to obtain the composition fluctuation parameters; Indexing to obtain the processing error parameters according to the processing parameters, including: Invoking a processing parameter library, where the processing parameter library is constructed based on historical processing parameters and includes a sample processing parameter set and a sample processing error parameter set, and each processing error parameter is the average error range when different sample processing parameters are controlled; Inputting the processing parameters into the processing parameter library and indexing to obtain the processing error parameters; Collecting the fluid characteristic parameters in the pump body operation scenario and analyzing to obtain the fluid complexity coefficient and the operation quality requirement coefficient, including: Collecting the fluid parameters in the pump body operation scenario to obtain a fluid parameter sequence; Calculating the mean value of the fluid parameter sequence to obtain the average fluid parameter; Randomly selecting several fluid parameters within the fluid parameter sequence, calculating the absolute error ranges from the average fluid parameter respectively, and calculating the mean value to obtain the fluid complexity coefficient; Taking the average fluid parameter as the operation quality requirement coefficient; Configuring stepwise prediction resources according to the composition fluctuation parameters, processing error parameters, fluid complexity coefficient, and operation quality requirement coefficient, and performing mechanical property prediction of the casting pump body to obtain the mechanical property parameters of the pump body.

2. The method for predicting the mechanical properties of a casting pump body based on processing parameters according to claim 1, wherein Processing the historical composition parameter sequence to obtain the composition fluctuation parameters, including: Calculating the mean value of the historical composition parameter sequence to obtain the average historical composition parameter; Randomly selecting several historical composition parameters within the historical composition parameter sequence, calculating the absolute error ranges from the average historical composition parameter respectively to obtain multiple composition absolute error ranges; Calculating the mean value of the multiple composition absolute error ranges to obtain the composition fluctuation parameters.

3. The method for predicting the mechanical properties of a casting pump body based on processing parameters according to claim 1, characterized in that, Configuring stepwise prediction resources according to the composition fluctuation parameters, processing error parameters, fluid complexity coefficient, and operation quality requirement coefficient, and performing mechanical property prediction of the casting pump body to obtain the mechanical property parameters of the pump body, including: Training an array of stepwise pump body performance predictors, where the array of stepwise pump body performance predictors includes multiple pump body performance predictors corresponding to multiple performance prediction coefficient intervals, and the training data volume of each pump body performance predictor is different; Calculating the performance prediction coefficient according to the composition fluctuation parameters, processing error parameters, fluid complexity coefficient, and operation quality requirement coefficient; Selecting the corresponding pump body performance predictor according to the performance prediction coefficient interval in which the performance prediction coefficient falls, inputting the processing parameters, and predicting and outputting to obtain the mechanical property parameters of the pump body.

4. The method for predicting the mechanical properties of a casting pump body based on processing parameters according to claim 3, wherein Training an array of stepwise pump body performance predictors, including: Collecting a sample processing parameter set and a sample mechanical property parameter set according to the pump body casting data within historical time; Using machine learning to construct multiple pump body performance predictors; Partition the set of sample processing parameters and the set of sample mechanical property parameters to obtain multiple sets of pump body performance prediction training data with different data volumes, and the data volumes of the multiple sets of pump body performance prediction training data are distributed from large to small; Respectively use the multiple sets of pump body performance prediction training data to conduct supervised training and testing on the multiple pump body performance predictors until convergence; Obtain multiple performance prediction coefficient intervals, where the performance prediction coefficient is greater than or equal to 0 and less than or equal to 1; Construct a mapping relationship with the multiple performance prediction coefficient intervals according to the training data volumes of the multiple pump body performance predictors, where the size of the performance prediction coefficient interval is positively correlated with the training data volume of the pump body performance predictor.

5. The method for predicting the mechanical properties of a casting pump body based on processing parameters according to claim 3, wherein Calculate the performance prediction coefficient according to the composition fluctuation parameter, processing error parameter, fluid complexity coefficient and operation quality requirement coefficient, including: According to the historical data of pump body casting, collect the maximum composition fluctuation parameter, maximum processing error parameter, maximum fluid complexity coefficient and maximum operation quality requirement coefficient within the historical time; Calculate the ratios of the composition fluctuation parameter, processing error parameter, fluid complexity coefficient and operation quality requirement coefficient to the maximum composition fluctuation parameter, maximum processing error parameter, maximum fluid complexity coefficient and maximum operation quality requirement coefficient respectively, and calculate the performance prediction coefficient by weighted calculation.

6. A system for predicting the mechanical properties of a casting pump body based on processing parameters, characterized in that, The system is used to execute the method for predicting the mechanical properties of a cast pump body based on processing parameters according to any one of claims 1-5, including: A processing parameter acquisition module, configured to acquire the processing parameters for current pump body casting and the composition fluctuation parameters for casting within a recently preset time range; A processing error parameter indexing module, configured to index and obtain the processing error parameter according to the processing parameter; A fluid characteristic parameter analysis module, configured to acquire the fluid characteristic parameters in the pump body operation scenario and analyze to obtain the fluid complexity coefficient and the operation quality requirement coefficient; A mechanical property prediction module, configured to configure stepped prediction resources according to the composition fluctuation parameter, processing error parameter, fluid complexity coefficient and operation quality requirement coefficient, conduct prediction on the mechanical properties of the cast pump body, and predict the mechanical property parameters of the pump body.

Images