WO2020228025A1

WO2020228025A1 - Method and apparatus for performing modeling analysis on data of type of machine tools

Info

Publication number: WO2020228025A1
Application number: PCT/CN2019/087284
Authority: WO
Inventors: 罗章维; 俞悦; 冯程; 曲颖; 王焦剑; 施内加斯丹尼尔
Original assignee: 西门子股份公司; 西门子（中国）有限公司
Priority date: 2019-05-16
Filing date: 2019-05-16
Publication date: 2020-11-19

Abstract

Provided are a method and an apparatus for performing modeling analysis on data of a type of machine tools. The method comprises: acquiring first data of multiple parameters of each of multiple machine tools which are of the same type; constraining, according to a pre-determined rule, at least one component of a first mathematical model by using at least one unknown coefficient, the first mathematical model representing a relationship between the multiple parameters; determining, on the basis of the acquired first data of the multiple parameters of each of the machine tools and the constrained first mathematical model, second data representing a fitting level of the first mathematical model for the type of the machine tools; determining, on the basis of the second data, whether the first mathematical model fits; and if the first mathematical model is determined as not fitting, acquiring a new first mathematical model. In this way, the invention provides an analysis model better fitting a specific type of machine tools, thereby enabling acquisition of an accurate data analysis result.

Description

Method and equipment for modeling and analyzing data of one type of mechanical tool

Technical field

The invention relates to the field of monitoring, in particular to monitoring/analyzing the performance of mechanical tools.

Background technique

In an industrial system, the performance of mechanical tools such as motors, frequency converters, gearboxes, etc. play a very important role in the normal operation of the entire industrial system. Timely monitoring of the status of these mechanical tools, fault diagnosis and early warning of them, or prediction of their service life, it is possible to predict in time before the mechanical tools actually fail or reach their service life, so as to ensure these Machine tools operate with good performance. Usually, the above-mentioned purpose is achieved by monitoring/analyzing the data from the machine tool. For example, it is possible to predict the remaining service life of the machine tool by analyzing the time-varying law of the parameters related to its performance obtained from the machine tool. .

According to existing methods, a long and short memory neural network (Long short memory network) can be used to predict the remaining service life of mechanical tools. This method needs to know all the data of the machine tool from the beginning of use to the end of its life. Another method is to perform regression analysis by fitting the known data of machine tool parameters to a certain curve equation. It is often difficult to determine what type of curve equation should be used for a specific type of machine tool to obtain the most accurate fit.

It is desirable to provide improved monitoring/analysis of data execution on machine tools.

Summary of the invention

It is expected to provide a method and equipment for modeling and analysis of machine tool data, which can provide a more suitable analysis model for a type of machine tool, and obtain more accurate data analysis results, such as more accurate execution of fault warning and/or Life prediction.

According to one aspect, a method for modeling and analyzing data of a type of mechanical tool is provided. The method includes obtaining first data of a plurality of parameters of each of the plurality of mechanical tools of the type; using at least one unknown coefficient according to a predefined rule to restrict at least one element of the first mathematical model, The first mathematical model represents the relationship between the plurality of parameters; based on the acquired first data of the plurality of parameters of each machine tool and the constrained first mathematical model, it is determined to represent the Second data on the suitability of the first mathematical model to the type of machine tool; determine whether the first mathematical model is suitable based on the second data; and if it is determined that the first mathematical model is not suitable, obtain The new first mathematical model.

According to another aspect, a device for modeling and analyzing data of one type of machine tool is provided. The device includes an obtaining unit, which is used to obtain the first data of a plurality of parameters of each of the plurality of mechanical tools of the type; a constraint unit, which is used to use at least one pair of unknown coefficients according to a predefined rule At least one component element of the first mathematical model is constrained, the first mathematical model represents the relationship between the plurality of parameters; and a determining unit, which is based on the acquired parameters of the plurality of parameters of each machine tool The first data and the constrained first mathematical model determine second data representing the suitability of the first mathematical model to the type of machine tool, and determine the first mathematical model based on the second data Whether the model is suitable; wherein, if the determining unit determines that the first mathematical model is not suitable, the acquiring unit acquires a new first mathematical model.

According to another aspect, a machine-readable medium is provided, and program instructions are stored, which when executed by a processor are used to perform the method according to the various embodiments of the present invention.

According to another aspect, a modeling analysis system is provided, including a memory that stores program instructions; and a processor that runs the program instructions to execute the method according to various embodiments of the present invention.

According to any aspect of the present invention, the parameter data from a plurality of mechanical tools of the same type is used to check whether the current mathematical model is suitable for modeling the current type of mechanical tool, and when it is determined that it is not suitable, continue to check the next mathematical model , Which can find a more accurate mathematical model for modeling current types of mechanical tools.

According to an embodiment of any aspect of the present invention, the new first mathematical model is determined as the current first mathematical model, and the constraint is repeatedly executed for the current first mathematical model, and the second mathematical model is determined The data and the determination are appropriate.

According to this embodiment, when it is determined that the previous first mathematical model is not suitable for modeling the current type of mechanical tool, it can be further judged whether the new first mathematical model is suitable. As a result, iteratively find a more suitable mathematical model for this type of machine tool.

According to another embodiment of any aspect of the present invention, constraining at least one component element of the first mathematical model according to a predefined rule includes at least for each independent variable in the first mathematical model, using the The first unknown coefficient in the at least one unknown coefficient is weighted and constrained and/or the second unknown coefficient in the at least one unknown coefficient is used for scaling.

According to this embodiment, weighting constraints on each independent variable in the first mathematical model can ensure that the dimensions of the equation on the left and right of the mathematical model are consistent, and the independent variables are scaled, taking into account the manufacturer’s The difference in scale.

According to another embodiment of any aspect of the present invention, the determining the second data includes determining, for each mechanical tool, based on the first data of the plurality of parameters of the mechanical tool, The at least one unknown coefficient of the constrained first mathematical model of the machine tool, thereby obtaining the second mathematical model of the machine tool; and the first parameter based on the plurality of parameters of each machine tool Data and the second mathematical model of each machine tool to determine the second data.

According to this embodiment, the current first mathematical model is first fitted to the first data of each mechanical tool, thereby determining the second mathematical model for each mechanical tool, and then according to the second mathematical model and each mechanical tool. The first data of each machine tool is used to determine the second data, which considers the data of each machine tool to determine the second data indicating the suitability of the current first mathematical model to the current type of machine tool.

According to a further embodiment of any aspect of the present invention, the determining the second data includes, for each machine tool, the first data and the machine tool based on the plurality of parameters of the machine tool The second mathematical model of the tool determines the loss data for the mechanical tool, and the loss data represents the output data of the selected parameter determined according to the second mathematical model of each mechanical tool and the acquired selected The difference between the first data of the parameters; determining the sum of a plurality of loss data for the plurality of machine tools; and determining whether the first mathematical model is suitable based on the sum of the plurality of loss data.

According to this embodiment, it is determined whether the current first mathematical model is suitable based on the loss data obtained by modeling the first data of each mechanical tool using the current first mathematical model, which is to determine whether the current first mathematical model is suitable It is used to model the current type of mechanical tools to provide a basis for judgment.

According to another embodiment of any aspect of the present invention, the first mathematical model is determined by using symbolic regression based on genetic algorithms; the method further includes when it is determined that the first mathematical model is not suitable, based on the The sum of a plurality of loss data determines the new first mathematical model through symbolic regression based on genetic algorithm.

According to this embodiment, after determining that the current first mathematical model is not suitable, the sum of loss data indicating the suitability of the current first mathematical model can be used as feedback and returned to the symbolic regression based on genetic algorithm to further optimize the mathematical model. Model, and thereby obtain the optimized mathematical model as the new first mathematical model, which increases the convergence speed of the entire algorithm and obtains the most suitable mathematical model faster.

According to another embodiment of any aspect of the present invention, the method further includes outputting the first mathematical model and the constrained first mathematical model for the type of machine tool when it is determined that the first mathematical model is suitable And/or a second mathematical model for each of the at least one target machine tool included in the plurality of machine tools.

According to this embodiment, not only can a mathematical model that is more suitable for modeling certain parameter data of this type of machine tool be obtained; when a plurality of machine tools include at least one target machine tool, it can also be output for at least one target at one time. At least one fitted mathematical model of the machine tool to provide accurate monitoring/analysis of the at least one target machine tool.

Description of the drawings

Hereinafter, the above-mentioned characteristics, technical characteristics, advantages and implementation methods of the present invention will be further described through the description of the preferred embodiments in a clear and easy-to-understand manner in conjunction with the following drawings.

Figure 1 shows a flow chart of a method for modeling and analyzing data of a type of mechanical tool according to an embodiment of the present invention;

Figure 2 shows the first data of a plurality of parameters obtained from seven numerically controlled machine tools according to an embodiment of the present invention;

3A-3D show curves of a second mathematical model of a target mechanical tool acquired according to an embodiment of the present invention;

Figure 4 shows a flowchart of a method for modeling and analyzing data of a type of mechanical tool according to another embodiment of the present invention;

FIG. 5 shows a flowchart of a method for modeling and analyzing data of a type of mechanical tool according to another embodiment of the present invention;

Figure 6 shows a block diagram of a device for modeling and analyzing data of a type of machine tool according to an embodiment of the present invention;

Fig. 7 shows a modeling analysis system according to an embodiment of the present invention.

The various aspects and features of various embodiments of the present invention are described with reference to the above-mentioned drawings. The above drawings are only schematic and not restrictive. Without departing from the gist of the present invention, the size, shape, label, or appearance of the various elements in the above drawings may be changed, and they are not limited to only those shown in the drawings in the specification.

Reference mark list

100, 200, 300 Methods for modeling analysis

110, 210, 310 Get the first data of multiple parameters

120, 220, 320 Obtain the first mathematical model

130, 230, 330 Constraint

140, 240, 340 Determine the second data

150, 250, 350 Determine whether it is appropriate

160, 260, 360 output

341 Determine the second mathematical model

342 Determine whether a predetermined number of first mathematical models have been obtained

10 Equipment used for modeling analysis

11 Acquisition unit

12 Constraint unit

13 Determine the unit

14 Output unit

15 Evaluation unit

20 Modeling analysis system

21 Memory

22 processor

Detailed ways

Usually, regression analysis is performed on the known parameter data from a specific mechanical tool, and a specific curve model is established. According to the model, the service life of the specific mechanical tool or the early warning of its failure can be predicted. It is unknown what form of model can be used in such regression analysis to obtain more accurate prediction results; and because the model is built only based on the known parameter data of a specific mechanical tool, it is difficult to ensure the accuracy of the model.

According to various embodiments of the present invention, on the one hand, different forms of models are verified to find the model that most accurately models the current type of machine tool; on the other hand, multiple models of the same type of machine tool are used. Known parameter data is used to verify the suitability/accuracy of modeling of this type of machine tool, thereby enabling more accurate modeling and analysis of this type of machine tool data and providing more accurate faults Early warning and/or life prediction.

Fig. 1 shows a flowchart of a method 100 for modeling and analyzing data of a type of mechanical tool according to an embodiment of the present invention. According to the embodiment of the present invention, it is possible to simultaneously perform modeling analysis on data of a plurality of mechanical tools of the same type. It can be expected that only a part of the plurality of mechanical tools or one mechanical tool is the target mechanical tool of interest, and the data of other mechanical tools of the same type is introduced only to find the mathematical model form for the target mechanical tool more accurately. It can also be expected that the plurality of mechanical tools does not contain the target mechanical tool, but only to find the most suitable mathematical model for the type of mechanical tool. After the mathematical model is found, the mathematical model can be fitted to the target of the type. Machine tool data. It can also be expected that the plurality of machine tools are all target machine tools, and thus, the respective fitting mathematical models for the plurality of machine tools of the same type can be found at one time.

According to the method 100, at 110, first data of a plurality of parameters of each of the plurality of machine tools is acquired, where the plurality of machine tools are the same type of machine tool, preferably, the so-called same type does not only refer to These mechanical tools belong to the same type from the point of view of their functions/functions, which also means that these mechanical tools are all operated in the same or similar operating environment. For example, they are the same type of motors working in the same environment. The multiple parameters of these mechanical tools are related to their performance. For example, for a motor, its operating temperature may indicate its performance and aging trend. Therefore, the operating temperature data of the motor can be obtained over time. In this case, multiple parameters can refer to the working temperature of the motor and the corresponding time. Similarly, in a manufacturing plant that uses CNC machine tools to process workpieces, considering that the performance of CNC machine tools will deteriorate over time, and the parameters of CNC machine tools such as temperature will rise to a certain threshold as their performance deteriorates, this may indicate Some parts need to be replaced. Therefore, the plural parameters are the temperature of the CNC machine tool and the corresponding time.

Or, in a manufacturing plant that uses chillers to maintain the production temperature at a predetermined level, considering the environment and plant infrastructure, in order to maintain the predetermined temperature, the chiller needs to reach a certain cooling load, which means that each chiller needs Consuming a certain amount of energy, it is desirable to model the energy consumed by the chiller and the corresponding cooling load to predict the energy required to reach the predetermined cooling load or the cooling load that can be achieved under the predetermined energy. In this case, multiple parameters can be the energy consumed by the chiller and the corresponding cooling load.

The first data of a plurality of parameters for each machine tool can be obtained separately, and they can be stored for processing. It is expected that the first data of a plurality of parameters of each machine tool can be obtained through various types of sensors installed to the machine tool.

Figure 2 shows the first data obtained from seven CNC machine tools, where the x-axis represents time and the y-axis represents the reciprocal of the energy used. According to this embodiment, the energy used by this type of machine tool is modeled over time.

At 120, a first mathematical model is obtained, which can represent the relationship between a plurality of parameters related to the performance of each mechanical tool. For example, for motors or CNC machine tools, it can represent the relationship between temperature and time, or the relationship between used energy and time. Although the first mathematical model is used here, this does not mean any limitation, and is only for distinguishing from the second mathematical model later.

For example, the first mathematical model obtained can be

y=exp(tanh(-t))+t ² (1)

Among them, t represents time and y represents energy used. The first mathematical model can be obtained in a variety of ways. In one embodiment, a plurality of expected mathematical models can be stored in a memory, and one of the mathematical models can be obtained from the memory as the first mathematical model. In a preferred embodiment, the first mathematical model can be a mathematical model determined by using a symbolic regression method based on genetic algorithms. This will be described in more detail in the following embodiments.

The mathematical model can be selected according to the object to be monitored/analyzed. For example, when the object to be monitored/analyzed does not involve periodic motion, for example, when the vibration parameters of the machine tool are not monitored, it is not necessary to select operations such as sine and cosine These periodic mathematical models can speed up the convergence rate of the entire analysis.

At 130, at least one component element in the acquired first mathematical model is restricted. In one embodiment, only the structure of the acquired first mathematical model may be retained, and one or more of the constituent elements including independent variables, arithmetic functions, and/or constants may be respectively restricted. Preferably, also An offset can be added to the first mathematical model. At least one unknown coefficient may be used to perform one or more constraints on the first mathematical model according to a predefined rule. In a preferred embodiment, at least for each independent variable in the first mathematical model, a first unknown coefficient in at least one unknown coefficient is used for weighted constraint and/or a second unknown coefficient in at least one unknown coefficient is used for Scaled. The independent variables in the first mathematical model can be weighted and constrained to ensure that the dimensions of both sides of the equation of the mathematical model are consistent, and the independent variables can be scaled to consider the difference of each machine tool at the factory setting. Further preferably, the calculation function in the first mathematical model can be weighted and restricted to further ensure that the dimensions of both sides of the equation of the mathematical model are consistent.

For example, for a mathematical model representing sensor data y that changes with time t, the following constraint rules can be predefined:

Independent variable t→(m*t+c);

Operation function (such as exponential function exp, logarithmic function log, etc.) e→n*e

Constant → K

Mathematical model → Mathematical model + offset O

Among them, m, c, n, K, O are unknown coefficients. In the above-mentioned predefined constraint rules, m is a weighted constraint on the independent variable t, and c is a scaled on the independent variable t. According to the predefined rule, if the obtained first mathematical model is as shown in the above formula (1), each component element in the first mathematical model can be restricted to obtain the following constrained first mathematical model .

y=n ₁ *exp(n ₂ *tanh(-(m ₁ *t+c ₁ )))+(m ₂ *t+c ₂ ) ^K1 +O (2)

Among them, m ₁ , m ₂ , c ₁ , c ₂ , n ₁ , n ₂ , K ₁ , and O are all unknown coefficients.

As shown in the above formula (2), it is preferable to use different unknown coefficients m ₁ , m ₂ and c ₁ , c ₂ for each independent variable in the first mathematical model to perform weighting constraints and scaling. Although the first mathematical model shown in formulas (1) and (2) only involves the same independent variable t, which appears twice, it can also be expected that multiple independent variables will be involved in some mathematical models. In this case, each independent variable is expected to be weighted and constrained and scaled separately.

At 140, based on the above-mentioned constrained first mathematical model and the first data obtained at 110 from the plural parameters of each of the plural mechanical tools of the same type, it is determined to represent the current first mathematical model pair The second data of the suitability of this type of machine tool, in other words, the second data indicating the accuracy of modeling the first data of the plurality of parameters of this type of machine tool using the current first mathematical model .

Specifically, based on data of a plurality of parameters of each machine tool, at least one unknown coefficient for constraining the first mathematical model can be determined for each machine tool. This can be achieved by fitting the first data of a plurality of parameters from each machine tool based on the above-mentioned constrained first mathematical model, such as the above formula (2), to solve the above-mentioned constrained first mathematical model Unknown coefficient to achieve. As a result, a second mathematical model for each machine tool is obtained. In other words, the second mathematical model is a fitting mathematical model for each mechanical tool obtained by respectively fitting the constrained first mathematical model to the first data of each mechanical tool.

If data of multiple parameters from 7 mechanical tools of the same type are acquired at 110, the above constrained first mathematical model can be used to fit the parameter data of each of these 7 different mechanical tools , Thereby obtaining 7 curves for the 7 different mechanical tools, and then obtaining 7 second mathematical models, and each second mathematical model is for each of the 7 different mechanical tools.

For example, for the first mechanical tool, the following second mathematical model can be obtained:

y ₁ =n ₁₁ *exp(n ₂₁ *tanh(-(m ₁₁ *t+c ₁₁ )))+(m ₂₁ *t+c ₂₁ ) ^K11 +O ₁ (3).

For the second mechanical tool, the following second mathematical model can be obtained:

y ₂ =n ₁₂ *exp(n ₂₂ *tanh(-(m ₁₂ *t+c ₁₂ )))+(m ₂₂ *t+c ₂₂ ) ^K12 +O ₂ (4).

By analogy, for the i-th machine tool, the following second mathematical model can be obtained:

y _i ＝n _1i *exp(n _2i *tanh(-(m _1i *t+c _1i )))+(m _2i *t+c _2i ) ^K1i +O _i (i=1，2……)(5 ).

The above m _1i , m _2i , c _1i , c _2i , n _1i , n _2i , K _1i , O _i are all known coefficients derived by curve fitting for parameter data of different mechanical tools.

After the second mathematical model for each of the multiple mechanical tools of the same type has been determined, it can be determined based on the data of the multiple parameters of each mechanical tool and its corresponding second mathematical model to indicate the use of The first mathematical model obtained by 120 is the second data of suitability for modeling the type of mechanical tool.

In a specific embodiment, the loss for each mechanical tool can be determined, for example, by substituting the data of multiple parameters of each mechanical tool into the corresponding second mathematical model to obtain the parameter value derived from the second mathematical model The difference between the actual parameter value of the machine tool and the actual parameter value to determine the loss data for each machine tool.

For example, for a certain machine tool, the sensor data representing the temperature change over time is obtained, and after the second mathematical model (for example, the above formula (5)) for the machine tool is determined, the obtained time point can be taken with Enter the second mathematical model, derive the corresponding y value, and then compare the derived y value with the sensor data actually measured at the time point to obtain the loss data for the time point, and finally determine the loss for the mechanical tool data.

After the loss for each machine tool is determined, the total loss data for all machine tools of the same type can be determined. For example, the total loss data can be calculated by using the following formula to sum the loss data for each machine tool.

Among them, y _{i, data} is the sensor measurement data of the parameters of the machine tool acquired at 110.

Although the second data is described with reference to the total loss data, it can be expected to express the suitability of the first mathematical model in other forms.

After the second data is determined at 140, it can be determined at 150 whether the current first mathematical model is suitable based on the second data, that is, whether the current first mathematical model is suitable for modeling the current type of machine tool.

In a specific embodiment, whether the first mathematical model is suitable can be determined based on the above-mentioned total loss data. For example, the total loss data can be compared with a predetermined threshold range. If the total loss data is within the predetermined threshold range, it indicates that the first mathematical model is acceptable, that is, the form of the currently used first mathematical model is appropriate , No need to reacquire. Otherwise, it indicates that the first mathematical model needs to be reacquired.

In another embodiment, the total loss data determined for the previous first mathematical model can be compared with the total loss data determined according to the current first mathematical model, and whether the current first mathematical model is suitable can be determined according to the comparison result.

If it is determined at 150 that the current first mathematical model is not suitable for modeling the current type of mechanical tool, or the accuracy of modeling the current type of mechanical tool using the current first mathematical model does not meet the predetermined standard, you need to return 120 obtain a new first mathematical model to represent the relationship between a plurality of parameters related to the performance of a plurality of machine tools. For example, a new first mathematical model can be obtained from the memory. After that, the process from 130 to 150 is repeated for the new first mathematical model.

If it is determined at 150 that the current first mathematical model is suitable, it indicates that the accuracy of modeling the current type of mechanical tool using the current first mathematical model meets the predetermined standard. At this time, the first mathematical model or /The first mathematical model constrained so that the form of the first mathematical model is used when modeling the same type of machine tool later. In a preferred embodiment, when the plurality of machine tools includes the target machine tool, the second mathematical model for each machine tool in at least one machine tool as the target machine tool among the plurality of machine tools can be output , For further processing.

For example, the second mathematical model of each target machine tool can be used to predict the remaining service life of the machine tool. In this embodiment, a predetermined threshold for sensor data, such as temperature data, can be obtained, and the predetermined threshold represents parameter data corresponding to the end of the life of the machine tool, such as the temperature at the end of its life. According to the predetermined threshold and the determined second mathematical model as shown in formula (5), the time corresponding to when the sensor data _yi reaches the predetermined threshold can be determined, and the current time can be easily determined according to the determined time and the current time. The remaining service life of the machine tool. In another specific application, the second mathematical model can be used to predict the energy consumed by the chiller as described above to reach the predetermined cooling load, and vice versa.

In one embodiment, after the above-mentioned cyclic processing is performed on the plurality of first mathematical models, it is determined at 150 that the following constrained first mathematical model has the lowest loss data, or the loss data meets a predetermined standard,

y(t)=(a ₁ *t+b ₁ )*a ₂ *exp(-(a ₃ *t+b ₂ ))+a ₄ *exp(-(a ₅ *t+b ₃ ))+D (7) The constrained first mathematical model can be output. Among them, a ₁ , a ₂ , a ₃ , a ₄ , a ₅ , b ₁ , b ₂ , b ₃ and D are unknown coefficients.

In a further embodiment, the second mathematical model of the target machine tool obtained according to the first mathematical model can also be output.

If for the above-mentioned seven mechanical tools of the same type, four of them are target mechanical tools, according to the form of the first mathematical model as determined by formula (7), the fitted second mathematical models of these four mechanical tools are respectively for

y ₁ = -1.5*(2.3*t-2.4)*exp(-0.07*t+1)-0.1*exp(0.2*t-0.2)+53.8 (8)

y ₂ =1*10 ^-5 *(-0.2*t+0.3)*exp(0.5*t+0.3)-6*10 ^-4 *exp(0.5*t-1.3)+10.0

(9)

y ₃ =15.5*(-13.1*t+367.4)

*exp(0.9*t+27.2)-4.2*10 ^-9 *exp(0.9*t+0.3)+9.99 (10)

y ₄ ＝7.8*(-7.6*t+121.0)

*exp(-0.9*t+37.8)-5.4*10 ^-13 *exp(1.2*t-0.3)+9.99 (11)

At 160, the above-mentioned second mathematical models (8)-(11) can be output for further analysis of these target machine tools. It can be seen from the above second mathematical model (8)-(11) that there are some items with very small coefficients, such as ^10-13 . In regression analysis according to the prior art, these items with very small coefficients are often ignored. Obviously, using the above-mentioned method according to the embodiment of the present invention, a more accurate model can be obtained. 3A-3D show the curves of the above-mentioned second mathematical model. The x-axis represents time, and the y-axis represents the value related to sensor data, specifically the reciprocal of the energy used.

FIG. 4 shows a flowchart of a method 200 for modeling and analyzing data of a type of machine tool according to another embodiment of the present invention. In this embodiment, a symbolic regression method based on genetic algorithm is used to obtain the first mathematical model, and the first mathematical model is gradually optimized based on the second data determined above. Specifically, after acquiring the first data of the plural parameters of the plural mechanical tools at 210, at 220, for example, the first mathematical model is determined by using a symbolic regression method based on genetic algorithm based on the acquired data. At 230, the same processing as 130 is performed. After the second data representing the suitability of the current first mathematical model for modeling the type of machine tool is determined at 240, at 250, the second data for the current first mathematical model is compared with the previous first mathematical model. Compare the second data of the model, for example, calculate the difference between the second data for the current first mathematical model and the second data for the previous mathematical model, and it can also be expected to calculate the second data for the current first mathematical model and For the difference between the average values of the second data of all previous first mathematical models; based on the comparison result, it is determined whether the current first mathematical model is suitable. For example, the comparison result can be compared with a predetermined threshold. It is said that the above determined difference can be compared with a predetermined threshold. When the difference is less than the predetermined threshold, it indicates that as the first mathematical model is optimized using the symbolic regression method based on genetic algorithms, the current first mathematical model is used to The loss of parameter data modeling of the current type of mechanical tool is reduced in line with expectations. At this time, it can proceed to 260 to output the current first mathematical model, the constrained current first mathematical model, or the second mathematical model corresponding to the target mechanical tool; If the difference is still greater than the predetermined threshold, it means that the current first mathematical model still needs to be optimized, and the currently obtained second data is returned to 220, which is used in the symbolic regression method based on genetic algorithm to continue to optimize the first mathematical model.

FIG. 3 shows a flowchart of a method 300 for modeling and analyzing data of a type of machine tool according to another embodiment of the present invention. In the method 300, the processes 310-340 are the same as the processes 110-140 according to the method 100 shown in Fig. 1 described above. The difference is that after the second data is determined at 340, the second data corresponding to the smallest second data is determined by comparing the current second data with the second data determined for the previous first mathematical model at 341 Model, and store the second mathematical model corresponding to the smallest second data, and determine at 342 whether a predetermined number of first mathematical models have been repeatedly acquired, if it is determined that a predetermined number of first mathematical models have been acquired, then 360 directly outputs the first mathematical model corresponding to the current smallest second data or the second mathematical model for the target machine tool among the plurality of second mathematical models. And if it is determined in 342 that the predetermined number of first mathematical models have not been repeatedly obtained, then in 350, it is further determined whether the second data determined for the current first mathematical model meets a predetermined standard, for example, the same as that determined for the previous first mathematical model. Whether the difference of the second data is less than the predetermined threshold, if it is satisfied, output the current first mathematical model or the second mathematical model for the target machine tool from the plurality of second mathematical models determined according to the current first mathematical model at 360, if If not satisfied, return to 320 to obtain a new first mathematical model.

Although the method according to the present invention has been described with reference to the various embodiments shown in FIGS. 1-3, it can be understood that the various processes of the methods of the above various embodiments can be split/combined/changed to achieve corresponding effects, and The different processes of the various embodiments can also be split/recombined to form corresponding methods. The ultimate goal is to find the form of the mathematical model that most accurately models the parameter data of the current type of mechanical tool, and to find a specific mathematical model for each mechanical tool.

FIG. 4 shows a device 10 for modeling and analyzing data of a type of mechanical tool according to an embodiment of the present invention. The device 10 includes an acquisition unit 11, a restriction unit 12, a determination unit 13, and an output unit 14.

The obtaining unit 11 obtains data of a plurality of parameters of a plurality of mechanical tools of this type and the current first mathematical model. Optionally, the device 10 may include a memory for storing data of a plurality of parameters and a plurality of first mathematical models that may be used. The first mathematical model represents the relationship between a plurality of parameters related to the performance of this type of machine tool.

In one embodiment, the acquisition unit 11 has processing capability, which can determine the first mathematical model by using symbolic regression based on genetic algorithm, or it can also be expected to use symbolic regression based on genetic algorithm to determine the first mathematical model in other locations. The acquiring unit 11 only acquires the determined first mathematical model.

The restriction unit 12 receives the first mathematical model from the acquisition unit 11, and is configured to use at least one unknown coefficient to restrict at least one component element of the first mathematical model according to a predefined rule. Pre-defined rules can be stored in memory. The at least one component element refers to one or more of the independent variables, operation functions, constants, and/or offsets in the first mathematical model.

In one embodiment, the constraint unit 12 uses the first unknown coefficient in the at least one unknown coefficient for weighted constraint and/or uses the second unknown coefficient in the at least one unknown coefficient for each independent variable in the current first mathematical model. The coefficients are scaled.

The determining unit 13 receives the constrained first mathematical model from the constraining unit 12 and the first data of the plural parameters of the plural mechanical tools from the acquiring unit 11, based on the value of each of the plural mechanical tools of the same type. The first data of a plurality of parameters and the constrained first mathematical model determine the second data indicating the suitability of the current first mathematical model to the type of machine tool, and determine the current first mathematical model based on the second data Whether the model is suitable, that is, whether it is necessary to obtain a new first mathematical model. If the determining unit 13 determines that a new first mathematical model needs to be acquired, the acquiring unit 11 is notified to acquire the new first mathematical model as the current first mathematical model, which is further processed by the constraint unit 12 and the determining unit 13.

In a specific embodiment, for each of the plurality of machine tools, the determining unit 13 determines the first constrained mathematical model for the machine tool based on the first data of the plurality of parameters of the machine tool At least one unknown coefficient of, thereby obtaining a second mathematical model for the machine tool; and determining the second data based on the first data of the plurality of parameters of each machine tool and the corresponding second mathematical model. In one embodiment, the second data is loss data, which represents the difference between the output data of the selected parameter determined according to the second mathematical model of each machine tool and the acquired measurement value of the selected parameter The difference.

Specifically, for each machine tool, the determining unit 13 determines the loss data for the machine tool based on the first data and the second mathematical model of the plurality of parameters of the machine tool; determines the loss data for the plurality of machine tools The sum of the data obtains the total loss data; and based on the total loss data, it is determined whether the current first mathematical model is suitable.

As described above, in one embodiment, the first mathematical model is determined by using symbolic regression based on genetic algorithm; therefore, when the determining unit 13 determines that the first mathematical model is not suitable, the acquiring unit may use genetic algorithm-based Symbolic regression determines a new first mathematical model based on the sum of loss data. When the determining unit 13 determines that the first mathematical model is suitable, the output unit 14 is used to output the current first mathematical model, the constrained current first mathematical model, or the second mathematical model for the target machine tool among the plurality of machine tools. .

In a specific embodiment, the device 10 further includes an evaluation unit 15 for predicting the remaining service life of the corresponding mechanical tool based on the output second mathematical model for each mechanical tool.

It can be understood that the methods and devices of the various embodiments of the present disclosure can be implemented by computer programs/software. These software can be loaded into the working memory of the processor, and used to execute the method according to the embodiments of the present disclosure when running, thereby obtaining the equipment for modeling analysis.

Fig. 7 shows a modeling analysis system 20 according to an embodiment of the present invention. The modeling analysis system 20 includes a memory 21 and a processor 22. The memory 21 stores computer program instructions, which when executed, can implement the methods according to the embodiments of the present disclosure. The processor 22 is configured to run these program instructions to implement the method according to the embodiments of the present disclosure as described above. It can be understood that the memory 21 is in a remote location, such as the cloud, and the processor 22 receives program instructions from the cloud through the network.

Exemplary embodiments of the present disclosure cover both of the following: creating/using the computer program/software of the present disclosure from the beginning, and converting an existing program/software to the computer program/software using the present disclosure by means of updating.

According to another embodiment of the present disclosure, there is provided a machine (such as a computer) readable medium, such as a CD-ROM, wherein the readable medium has computer program code stored thereon, and the computer program code when executed The computer or the processor executes the method according to the embodiments of the present disclosure. The machine-readable medium is, for example, an optical storage medium or a solid-state medium supplied with or as part of other hardware.

The computer program for executing the method according to the various embodiments of the present disclosure may also be distributed in other forms, such as via the Internet or other wired or wireless telecommunication systems.

The computer program can also be provided on a network such as the World Wide Web, and can be downloaded from such a network to the working computer of the processor.

It must be pointed out that the embodiments of the present disclosure are described with reference to different subjects. In particular, some embodiments are described with reference to method-type claims, while other embodiments are described with reference to device/system-type claims. However, those skilled in the art will learn from the above and the following description, unless otherwise specified, in addition to any combination of features belonging to one type of subject matter, any combination of features related to different subjects is also deemed to be disclosed in this application Up. In addition, all the features can be combined to provide a synergistic effect greater than the simple addition of the features.

The specific embodiments of the present disclosure have been described above. Other embodiments are within the scope of the appended claims. In some cases, the actions or processing described in the claims may be performed in a different order than in the embodiments and still achieve desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or sequential order shown to achieve the desired result.

The present disclosure is described above with reference to specific embodiments, and those skilled in the art should understand that the technical solutions of the present disclosure can be implemented in various ways without departing from the spirit and basic characteristics of the present disclosure. The specific embodiments are merely illustrative and not restrictive. In addition, these embodiments can be combined arbitrarily to achieve the purpose of the present disclosure. The protection scope of the present disclosure is defined by the appended claims.

The term "comprising" in the specification and claims does not exclude the existence of other elements or processing, and expressions such as "first", "second" and the like do not denote order or limit the number. The function of each element described in the specification or described in the claims can also be divided or combined, and implemented by corresponding plural elements or a single element.

Claims

A method of modeling and analyzing the data of a type of machine tool (100, 200, 300), including

Acquiring first data of a plurality of parameters of each of the plurality of mechanical tools of the type;

Use at least one unknown coefficient to constrain at least one element of the first mathematical model according to a predefined rule (130, 230, 330), the first mathematical model representing the relationship between the plurality of parameters;

Based on the acquired first data of the plurality of parameters of each machine tool and the constrained first mathematical model, it is determined (140, 240, 340) to indicate that the first mathematical model affects the type of The second data of the suitability of the machine tool;

Determine (150, 250, 350) whether the first mathematical model is suitable based on the second data; and if it is determined that the first mathematical model is not suitable, obtain (120, 220, 320) a new first mathematical model .
The method (100, 200, 300) according to claim 1, further comprising

Determine the new first mathematical model as the current first mathematical model, and

For the current first mathematical model, the constraint is repeatedly executed, the determination of the second data and the determination of whether the first mathematical model is suitable.
The method (100, 200, 300) according to claim 1, wherein constraining at least one component element of the first mathematical model includes at least for each independent variable in the first mathematical model, using all The first unknown coefficient in the at least one unknown coefficient is weighted and restricted and/or the second unknown coefficient in the at least one unknown coefficient is used for scaling.
The method (100, 200, 300) according to claim 1, wherein said determining said second data comprises

For each machine tool, based on the first data of the plurality of parameters of the machine tool, determine (140, 240, 340) the constrained first mathematical model for the machine tool At least one unknown coefficient, thereby obtaining a second mathematical model of the machine tool; and

Based on the first data of the plurality of parameters of each machine tool and the second mathematical model of each machine tool, the second data (140, 240, 340) is determined.
The method (100, 200, 300) according to claim 4, wherein said determining said second data comprises

For each machine tool, based on the first data of the plurality of parameters of the machine tool and the second mathematical model of the machine tool, it is determined (140, 240, 340) for the machine tool Loss data of the tool, the loss data representing the difference between the output data of the selected parameter determined according to the second mathematical model of each mechanical tool and the acquired first data of the selected parameter;

Determine (140, 240, 340) the sum of a plurality of loss data for the plurality of machine tools; and

Based on the sum of the plurality of loss data, it is determined (150, 250, 350) whether the first mathematical model is suitable.
The method (100, 200, 300) according to claim 5, wherein the first mathematical model is determined by using symbolic regression based on genetic algorithms; the method further comprises

When it is determined that the first mathematical model is not suitable, based on the sum of the plurality of loss data, the new first mathematical model (120, 220, 320) is determined through symbolic regression based on genetic algorithm.
The method (100, 200, 300) according to any one of claims 4-6, further comprising

When it is determined that the first mathematical model is suitable, output (160, 260, 360) for the first mathematical model for the type of machine tool, the constrained first mathematical model and/or for the The second mathematical model of each of the at least one target machine tool among the plurality of machine tools is described.
Equipment (10) for modeling and analyzing the data of a type of mechanical tool, including

An obtaining unit (11), which is used to obtain first data of a plurality of parameters of each of the plurality of mechanical tools of the type;

A restriction unit (12), which is used to restrict at least one element of the first mathematical model using at least one unknown coefficient according to a predefined rule, the first mathematical model representing the relationship between the plurality of parameters; and

A determining unit (13), which determines, based on the acquired first data of the plurality of parameters of each machine tool and the constrained first mathematical model, determining the effect of the first mathematical model on the type Second data on the suitability of the machine tool, and determining whether the first mathematical model is suitable based on the second data;

Wherein, if the determining unit (13) determines that the first mathematical model is not suitable, the acquiring unit (11) acquires a new first mathematical model.
The device (10) according to claim 8, wherein:

The restriction unit (12) determines the new first mathematical model as the current first mathematical model to restrict it, and the determination unit (13) determines the current first mathematical model for the Second data and determine whether the current first mathematical model is suitable based on the second data.
The device (10) according to claim 8, wherein the constraint unit (12) uses the first unknown coefficient in the at least one unknown coefficient for at least each independent variable in the first mathematical model. Weighting constraints and/or using a second unknown coefficient among the at least one unknown coefficient for scaling.
The device (10) according to claim 8, wherein the determining unit (13) determines the specific value for each machine tool based on the first data of the plurality of parameters of the machine tool The at least one unknown coefficient of the constrained first mathematical model of the machine tool, thereby obtaining the second mathematical model of the machine tool; and the number of parameters based on the plurality of parameters of each machine tool The first data and the second mathematical model of each machine tool determine the second data.
The device (10) according to claim 11, wherein the determining unit (13)

For each machine tool, based on the first data of the plurality of parameters of the machine tool and the second mathematical model of the machine tool, determine the loss data for the machine tool, the loss data Represents the difference between the output data of the selected parameter determined according to the second mathematical model of each mechanical tool and the obtained value of the selected parameter;

Determine the sum of a plurality of loss data for the plurality of machine tools; and

Based on the sum of the plurality of loss data, it is determined (150, 250, 350) whether the first mathematical model is suitable.
The device (10) according to claim 12, wherein the first mathematical model is determined by using symbolic regression based on genetic algorithms; and when the determining unit (13) determines that the first mathematical model is not suitable When the time, the acquiring unit determines the new first mathematical model based on the sum of the plurality of loss data through symbolic regression based on the genetic algorithm.
The device (10) according to any one of claims 11-13, further comprising an output unit (14), when the determining unit (13) determines that the first mathematical model is suitable, the output unit (14) ) For outputting the first mathematical model for the type of machine tool, the constrained first mathematical model, and/or for each of the at least one target machine tool among the plurality of machine tools The second mathematical model of the machine tool.
A machine-readable medium storing program instructions, which when executed by a processor are used to execute the method according to any one of claims 1-7.
Modeling analysis system, including

Memory, which stores program instructions; and

A processor that runs the program instructions to execute the method according to any one of claims 1-7.