A parameter optimization method based on considering future running state of equipment

Abstract

The application relates to a parameter optimization method based on future operation state of equipment, which comprises the following steps: collecting historical state of equipment, input value data of operation parameters and target value data of operation state, forming a data set, firstly establishing a linear regression model between the input value data of operation parameters and the target value data of operation state, substituting the data set into the linear regression model to extract a relationship parameter, then predicting the target value data of operation state of the equipment at a future time point or time period, and finally constructing an equation group to predict the input value data of operation parameters of the equipment at a next operation state according to the predicted target value data of operation state. Through implementation of the application, the equipment can be efficiently operated according to the predicted operation state target, and the adverse influence of adjustment of operation parameters on the future equipment operation target in the equipment operation process is avoided, thereby providing a technical basis for efficient and stable operation and control of the equipment.

Description

Technical Field

[0001] This invention belongs to the fields of intelligent devices and data mining technology, and specifically relates to a parameter optimization method based on considering the future operating state of the device. Background Technology

[0002] Integrating next-generation information technologies to facilitate the digital and intelligent upgrading of equipment is a key breakthrough direction in the development plan of intelligent manufacturing, and is crucial for promoting high-quality equipment development and driving the integration of the digital economy and the real economy. Complex electromechanical equipment, as one of the main carriers of equipment digital and intelligent upgrading, suffers from incomplete clarity regarding the information and energy transfer mechanisms between its internal systems due to the complexity of the coupling. This results in input parameters failing to achieve the expected output or exhibiting inefficient performance.

[0003] Traditional methods for finding input parameters for complex electromechanical equipment mainly involve constructing a mechanistic model and then solving for the optimal input parameters based on experience or optimization algorithms. This method assumes that the theoretical complexity of the complex electromechanical equipment equals its actual complexity, and the obtained input parameters are empirical and static. Therefore, this method is prone to deviating from the actual operating state of complex electromechanical equipment and lacks transferability. Summary of the Invention

[0004] To optimize the impact of input parameter adjustments on the future operational goals of complex electromechanical equipment and address the shortcomings of traditional methods for finding input parameters that easily deviate from the actual operating state and lack transferability, this invention provides a parameter optimization method based on considering the future operating state of the equipment. The aim is to predict the input values ​​of the equipment's operating parameters for the next time period or point in time during equipment operation, thereby ensuring efficient equipment operation. This invention relies on big data technology to directly mine the relationship between input parameters and output goals during equipment operation, finding the optimal output target that the complex electromechanical equipment may achieve in the future, and ultimately predicting the parameters that should be input for the equipment's operation in the future. This method avoids constructing complex mechanistic models and the established data model has strong transferability.

[0005] The technical solution adopted by the present invention to solve the above-mentioned technical problems is as follows:

[0006] A parameter optimization method considering the future operating state of equipment includes the following steps:

[0007] Step 1: Collect historical data on the device's operating parameter inputs and target operating status values ​​to obtain a dataset;

[0008] Step 2: Establish a linear regression model between the input values ​​of the equipment's operating parameters and the target values ​​of its operating status. The linear regression model is as follows:

[0009] Y i =W*X i +b (1)

[0010] Among them, Y i X represents the target value of the i-th operating state of the device. i W represents a column vector consisting of all input operating parameters for the i-th operating state of the device, where W represents Y. i and X i The row vector in the relational parameters, b represents Y i and X i The intercept in the relational parameters;

[0011] Substitute the dataset into the linear regression model to extract the row vector W and intercept b from the relation parameters;

[0012] Step 3: Construct a time series model using an autoregressive algorithm. Substitute all the target values ​​of the operating state into the time series model to obtain the predicted value of the next target value of the operating state adjacent to the current target value. The time series model is as follows:

[0013] f(Y1,Y2,…Y t )=P_Y t+1 (2)

[0014] Where f represents the functional expression of the time series model, Y t P_Y represents the target value of the current running state. t+1 This represents the predicted value of the target value for the next running state.

[0015] Step 4: Construct a system of equations to predict the input values ​​of the operating parameters of the equipment in the next operating state. Use an optimization algorithm to solve the feasible solution of the system of equations. The feasible solution is the predicted value of the input values ​​of the operating parameters of the equipment in the next operating state. The system of equations is as follows:

[0016] P_Y t+1 =WX i+1 +b (3)

[0017] min(|P_Y t+1 -(W*P_X i+1 +b)|) (4)

[0018] K1≤X i+1 ≤K2 (5)

[0019] Among them, P_X i+1K1 represents the column vector composed of the minimum values ​​of each operating parameter and K2 represents the column vector composed of the maximum values ​​of each operating parameter. K1 represents the feasible solution of the system of equations, which is the predicted value of the operating parameter input value of the device in the next operating state.

[0020] Compared with the prior art, the present invention has the following beneficial effects:

[0021] This invention proposes a parameter optimization method based on considering the future operating state of equipment. This method predicts the target value of the equipment's operating state at a future point in time or over a period of time, then constructs a system of equations. Based on the predicted target value of the equipment's operating state, it predicts the input values ​​of the equipment's operating parameters for the next operating state, thereby effectively determining the optimal input values ​​of the equipment's operating parameters for the next operating state. This completes the parameter optimization of the equipment, enabling it to operate efficiently according to the predicted operating state target. It avoids the adverse effects of adjusting the operating parameters during equipment operation on the future operating target of the equipment, providing a technical foundation for the efficient and stable operation and control of the equipment. Attached Figure Description

[0022] Figure 1 This is an overall flowchart of a parameter optimization method based on considering the future operating state of the equipment according to one embodiment of the present invention;

[0023] Figure 2 The prediction effect diagram of the linear regression model constructed in the embodiment of the present invention;

[0024] Figure 3 The prediction effect diagram of the time series model constructed in the embodiment of the present invention;

[0025] Figure 4 The diagram illustrates the convergence process of using a genetic algorithm to solve for the input values ​​of the device's operating parameters in the next operating state. Detailed Implementation

[0026] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0027] In one embodiment, the present invention proposes a parameter optimization method based on considering the future operating state of the equipment, which mainly includes the following steps: establishing the relationship between the operating parameters input by the equipment and the operating target of the equipment, predicting the operating target of the equipment at a future point in time (or time period) (each point in time or time period for completing a target is a running state), and predicting the input operating parameters under the next operating target of the equipment.

[0028] Specifically, such as Figure 1 As shown, the parameter optimization method provided by the present invention, which considers the future operating state of the equipment, includes the following steps:

[0029] Step 1: Collect the input values ​​of the device's operating parameters and the target values ​​of the operating status under historical conditions to obtain a dataset.

[0030] Step 2: The relationship between the input values ​​of the equipment's operating parameters and the target values ​​of its operating status is established using a linear regression model. This model is based on historical data (X...). i ,Y i (W,b) is obtained, and the linear regression model is shown in equation (1):

[0031] Y i =W*X i +b (1)

[0032] Among them, Y i X represents the target value of the i-th operating state of the device. i W represents a column vector consisting of all input operating parameters for the i-th operating state of the device, where W represents Y. i and X i The row vector in the relational parameters, b represents Y i and X i The intercept in the relational parameters.

[0033] Substitute the dataset into the linear regression model shown in formula (1) to extract the relational parameters—row vector W and intercept b.

[0034] Step 3: The operating target of the equipment at a future point in time (or time period) is predicted using an autoregressive model. The target is to predict the output target of the equipment in the next state. Specifically, a time series model is constructed using an autoregressive algorithm. All the operating state target value data are substituted into the time series model to obtain the predicted value of the next operating state target value adjacent to the current operating state target value. The time series model is shown in Equation (2):

[0035] f(Y1,Y2,…Y t )=P_Y t+1 (2)

[0036] Where f represents the functional expression of the time series model, Y t P_Y represents the target value of the current running state. t+1 This represents the predicted value of the next operating state target value, which is the target value of the next operating state predicted based on the target values ​​of the current state and the t-1 preceding operating states.

[0037] Optionally, when constructing a time series model using an autoregressive algorithm, the model order can be set to 1 to simplify calculations.

[0038] Step 4: Construct a system of equations to predict the input values ​​of the operating parameters of the equipment in the next operating state. Use an optimization algorithm to solve the feasible solution of the system of equations. This feasible solution is the predicted value of the input values ​​of the operating parameters of the equipment in the next operating state. The system of equations is as follows:

[0039] P_Y t+1 =WX i+1 +b (3)

[0040] min(|P_Y t+1 -(W*P_X i+1 +b)|) (4)

[0041] K1≤X i+1 ≤K2 (5)

[0042] Among them, P_X i+1 K1 represents the column vector composed of the minimum values ​​of each operating parameter and K2 represents the column vector composed of the maximum values ​​of each operating parameter. K1 represents the feasible solution of the system of equations, which is the predicted value of the operating parameter input value of the device in the next operating state.

[0043] After steps 1 to 3 above, P_Y t+1 Since W and b are known, the prediction of the operating parameter input values ​​of the device in the next operating state is to obtain the feasible solution P_X that satisfies the conditions according to equation (3). i+1 The solution process uses optimization algorithms such as genetic algorithms. The target value is shown in equation (4), and the constraints on the input values ​​of the operating parameters of the equipment are shown in equation (5).

[0044] The parameter optimization method proposed in this embodiment, which considers the future operating state of the equipment, predicts the target value of the equipment's operating state at a future point in time or over a period of time, then constructs a system of equations. Based on the predicted target value of the equipment's operating state, the method predicts the input values ​​of the equipment's operating parameters for the next operating state, thereby effectively determining the optimal input values ​​of the equipment's operating parameters for the next operating state. This completes the parameter optimization of the equipment, enabling it to operate efficiently according to the predicted operating state target. It avoids the adverse effects of adjusting the operating parameters during equipment operation on the future operating target of the equipment, providing a technical foundation for the efficient and stable operation and control of the equipment.

[0045] Next, taking the prediction of input parameters of the mechanical loading spindle, a key component of the high-temperature biaxial material mechanical property testing equipment, as an example, the process of parameter optimization method of the present invention will be further explained in detail.

[0046] The operation process of the mechanical loading spindle X-axis of the test equipment is selected. The input values ​​of the operation parameters are the X1 force value and X2 force value loaded in the positive and negative directions of the X-axis. The output target value of the operation state is the absolute value of the X coordinate of the center point of the X-axis. In this embodiment, the time span of the operation state is set to 10s, that is, the input parameters used every 10 seconds determine the operation state of the mechanical loading spindle X-axis.

[0047] The data of the mechanical loading spindle's operating parameters (X1 force value, X2 force value) and the output target value of the operating status (absolute value of the X coordinate of the X-axis center point) were collected at 10-second intervals to form a dataset with 259 samples.

[0048] The first 249 samples of the dataset were selected and divided into training and prediction data in a 4:1 ratio. A linear regression (LR) algorithm was then used to construct a linear regression model between the input parameters and the target values ​​of the running state. The prediction performance of the linear regression model, validated using 50 prediction datasets, is shown below. Figure 2 The mean absolute error is 0.058, which shows that the constructed linear regression model has excellent predictive performance.

[0049] Extract the relationship parameters between the parameter data and the target data in the linear regression model, namely W = [2.4031, -2.2067] and b = 1.2386 in equation (1).

[0050] All target values ​​of operating states were selected from the dataset. An autoregressive (AR) algorithm was used, with the model order set to 1, to construct a time series model of the target data. This model was then used to predict the target value of the mechanically loaded spindle in its next operating state. Based on the previous target value, predictions of the next operating state target value were performed 258 times in the dataset. The prediction performance of the time series model is compared below. Figure 3 The mean absolute error is 0.0101, which shows that the constructed time series model has excellent prediction performance.

[0051] Select the 250th operational state target value P_Y predicted using the constructed time series model. 250 =6.02292, which is P_Y in equation (3) t+1 Because the force values ​​X1 and X2 show an increasing trend over time, the true values ​​of the force values ​​X1 and X2 in the previous state of the predicted target value are selected as K1. Therefore, the K1 value relative to the 250th running state is [16.20483, 15.50293]. T The value of K2 is shown in equation (6), where n is the dataset that has been collected. In this embodiment, the value of n is 249, so the value of K2 is [16.72363, 15.99121]. T .

[0052]

[0053] For the parameter prediction of the 250th position state, referring to equations (4) and (5), based on the W value, b value, and P_Y obtained above... 250 Given the values ​​of K1 and K2, a representative optimization algorithm—the genetic algorithm—is used to solve for P_X. i+1 The value is encoded in binary, with a crossover probability of 0.4 and a mutation probability of 0.05. The fitness function f is shown in equation (7). Taking the next running state as an example, the parameter solution process is shown in [reference]. Figure 4 As the number of iterations increases, the target value shown in formula (4) approaches zero (the optimal target value is 0), and the convergence effect is good. The predicted and actual values ​​of force X1 are 32723.372 and 32714.84375, respectively, and the predicted and actual values ​​of force X2 are 31300.44596 and 31311.03516, respectively. It can be seen that the predicted values ​​of force X1 and force X2 are close to their respective actual values.

[0054]

[0055] Similar to the process of obtaining the parameter data for the 250th running state, the actual and predicted values ​​of the parameters for the 250th to 259th running states are shown in Table 1, which verifies the effectiveness of the parameter optimization method in this invention.

[0056] Table 1

[0057]

[0058]

[0059] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0060] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this invention patent should be determined by the appended claims.

Claims

1. A parameter optimization method considering the future operating state of equipment, characterized in that, Includes the following steps: Step 1: Collect historical data on the device's operating parameter inputs and target operating status values ​​to obtain a dataset; Step 2: Establish a linear regression model between the input values ​​of the equipment's operating parameters and the target values ​​of its operating status. The linear regression model is as follows: Y i =W*X i +b (1) Among them, Y i X represents the target value of the i-th operating state of the device. i W represents a column vector consisting of all input operating parameters for the i-th operating state of the device, where W represents Y. i and X i The row vector in the relational parameters, b represents Y i and X i The intercept in the relational parameters; Substitute the dataset into the linear regression model to extract the row vector W and intercept b from the relation parameters; Step 3: Construct a time series model using an autoregressive algorithm. Substitute all the target values ​​of the operating state into the time series model to obtain the predicted value of the next target value of the operating state adjacent to the current target value. The time series model is as follows: f(Y1,Y2,…Y t )=P_Y t+1 (2) Where f represents the functional expression of the time series model, Y t P_Y represents the target value of the current running state. t+1 This represents the predicted value of the target value for the next running state. Step 4: Construct a system of equations to predict the input values ​​of the operating parameters of the equipment in the next operating state. Use an optimization algorithm to solve the feasible solution of the system of equations. The feasible solution is the predicted value of the input values ​​of the operating parameters of the equipment in the next operating state. The system of equations is as follows: P_Y t+1 =WX i+1 +b (3) min(|P_Y t+1 -(W*P_X i+1 +b)|) (4) K1≤X i+1 ≤K2 (5) Among them, P_X i+1 K1 represents the column vector composed of the minimum values ​​of each operating parameter and K2 represents the column vector composed of the maximum values ​​of each operating parameter. K1 represents the feasible solution of the system of equations, which is the predicted value of the operating parameter input value of the device in the next operating state.

2. The parameter optimization method based on considering the future operating state of the equipment according to claim 1, characterized in that, When constructing a time series model using the autoregressive algorithm, the model order is set to 1.

3. A parameter optimization method based on considering the future operating state of equipment according to claim 1 or 2, characterized in that, The optimization algorithm used is a genetic algorithm.

Images