Optimization calculation method

Abstract

To provide an optimization calculation method capable of improving control accuracy even in a case where a calculation load is high.SOLUTION: An optimization calculation method determines a controlled variable of a plant by using nonlinear calculation in real time in a case where a calculation load is lower than a predetermined value, and determines the controlled variable of the plant by utilizing a calculation result of performing nonlinear calculation in the past in a case where the calculation load is higher than the predetermined value.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to an optimization computation method. [Background technology]

[0002] Patent Document 1 discloses a technique for reducing the computational load by setting a threshold value for a specific parameter and executing nonlinear optimization control when the parameter is in a certain direction relative to the threshold value. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2010-155528 A Summary of the Invention [Problem to be solved by the invention]

[0004] However, in the technology disclosed in Patent Document 1, in order to reduce the calculation load, when the parameters do not exceed a certain value, light calculations such as map lookup are performed for control, and the optimum solution at that time cannot be searched for, resulting in a decrease in the accuracy of the solution. Therefore, while nonlinear optimization control is not being performed, the control accuracy decreases.

[0005] The present invention has been made in consideration of the above problems, and an object of the present invention is to provide an optimization calculation method that can improve control accuracy even when the calculation load is high. [Means for solving the problem]

[0006] In order to solve the above-mentioned problems and achieve the object, the optimization calculation method according to the present invention is characterized in that, when the calculation load is lower than a predetermined value, a control variable of a plant is determined in real time using nonlinear calculation, and, when the calculation load is higher than the predetermined value, a control variable of the plant is determined using a calculation result of a nonlinear calculation performed in the past. Effect of the Invention

[0007] The optimization calculation method according to the present invention makes it possible to control a plant using the nonlinear calculation results even when the calculation load is high, and therefore has the effect of improving control accuracy. [Brief description of the drawings]

[0008] [Figure 1] FIG. 1 is a diagram showing an example in which the optimization calculation method according to the embodiment is applied to a system having only one calculation device. [Diagram 2] FIG. 2 is a diagram showing how the optimization target range is divided when there are multiple arithmetic units. [Diagram 3] FIG. 3 is a flowchart showing an example of calculation control for performing parallel calculations of nonlinear optimization calculations by a plurality of calculation devices. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0009] Hereinafter, an embodiment of an optimization calculation method according to the present invention will be described, however, the present invention is not limited to the embodiment.

[0010] The optimization calculation method according to the embodiment aims to reduce the calculation load of optimization calculation involving repeated calculation and array calculation, and reduces the calculation load by utilizing one or more calculation devices provided in the vehicle or an external calculation device such as a cloud. The optimization calculation method according to the embodiment reduces the calculation load without reducing the accuracy of the solution of the optimization control. This is particularly effective for control using a nonlinear optimization problem. The optimization calculation method according to the embodiment differs between a case having multiple calculation devices and a case having only one calculation device, but is common in terms of distributing calculations that require heavy processing.

[0011] In the case where there is only one calculation device, when the calculation load is lower than a predetermined value and there is a margin in the processing capacity, a nonlinear optimization calculation is performed (A1 in FIG. 1). The calculation result of the nonlinear optimization calculation is stored in a memory function such as a map (A2 in FIG. 1). When there is a margin in the processing capacity, the calculation result of the nonlinear optimization calculation is adopted as a time when real-time calculation of the nonlinear optimization calculation is possible (Yes in A3 in FIG. 1), a more optimal solution is selected based on the value of the evaluation function (A4 in FIG. 1), and a control instruction value is output to the plant 100. The plant 100 is, for example, a power plant in a vehicle, which includes a transaxle (power transmission device) composed of a rotating electric machine (motor generator) and a gear mechanism, and an engine.

[0012] On the other hand, when the calculation load is higher than a predetermined value and the processing capacity of the calculation device is not sufficient, the nonlinear optimization calculation is not performed, and a linear optimization calculation is performed as a calculation that requires relatively light processing (A5 in FIG. 1). Also, when the processing capacity is not sufficient, a value stored in a memory function such as a map is used as a time when real-time calculation of the nonlinear optimization calculation is not possible (No in A3 in FIG. 1), and the solution (value) calculated by the linear optimization calculation is compared with the solution (value) stored in the memory function to select a more optimal solution based on the value of the evaluation function (A4 in FIG. 1), and a control instruction value is output to the plant 100.

[0013] When there are multiple arithmetic devices, the target range for optimization (area for performing optimization) is divided for each arithmetic device. For example, as shown in FIG. 2, when there are four arithmetic devices, ECU (Electronic Control Unit) 1 to ECU 4, the area for performing optimization is divided into ZONE 1 to ZONE 4 for each ECU 1 to ECU 4. In this way, the calculation load of each arithmetic device (ECU 1 to ECU 4) is made lower than a predetermined value, and nonlinear optimization calculation is performed in parallel in each arithmetic device (ECU 1 to ECU 4). Then, the calculation results of the nonlinear optimization calculation performed in parallel in each arithmetic device (ECU 1 to ECU 4) are compared, and the calculation result with the smallest evaluation function is selected as the optimal solution, and a control instruction value is output to the plant 100.

[0014] Fig. 3 is a flowchart showing an example of arithmetic control in which nonlinear optimization calculations are performed in parallel by multiple arithmetic devices (ECU1 to ECU4). In the flowchart shown in Fig. 3, four arithmetic devices, ECU1 to ECU4, are used as the multiple arithmetic devices, and the target region for optimization (region for optimization) is divided into four.

[0015] First, in the process of step S1, the target range (region in which optimization is performed) for each of ECU1 to ECU4 is divided into four. With this, the process of step S1 is completed, and the process proceeds to the process of step S11, the process of step S21, the process of step S31, and the process of step S41 at the same time, and the nonlinear optimization calculations are performed in parallel by ECU1 to ECU4.

[0016] Regarding the ECU1, in the process of step S11, the ECU1 performs a nonlinear optimization calculation to output the optimal operation amount u1. Then, the process of step S11 is completed, and the process proceeds to step S12. In the process of step S12, the ECU1 calculates the value of the evaluation function at the optimal operation amount u1. Then, the process of step S12 is completed, and the process proceeds to step S2.

[0017] Regarding ECU2, in the process of step S21, ECU2 performs a nonlinear optimization calculation to output the optimal manipulated variable u2. Then, the process of step S21 is completed, and the process proceeds to step S22. In the process of step S22, ECU2 calculates the value of the evaluation function at the optimal manipulated variable u2. Then, the process of step S22 is completed, and the process proceeds to step S2.

[0018] Regarding the ECU3, in the process of step S31, the ECU3 performs a nonlinear optimization calculation to output an optimal manipulated variable u3. Then, the process of step S31 is completed, and the process proceeds to step S32. In the process of step S32, the ECU3 calculates a value of an evaluation function at the optimal manipulated variable u3. Then, the process of step S32 is completed, and the process proceeds to step S2.

[0019] Regarding the ECU4, in the process of step S41, the ECU4 performs a nonlinear optimization calculation to output an optimal manipulated variable u4. Then, the process of step S41 is completed, and the process proceeds to step S42. In the process of step S42, the ECU4 calculates a value of an evaluation function at the optimal manipulated variable u4. Then, the process of step S42 is completed, and the process proceeds to step S2.

[0020] In the process of step S2, the values ​​of the evaluation functions calculated by ECU1 to ECU4 are compared, and the manipulated variable that is the minimum value is output as the final result to the plant 100. Then, the process of step S2 is completed, and the example of the computation control ends.

[0021] As described above, the optimization calculation method according to the embodiment determines the control amount of the plant 100 using nonlinear calculation in real time when the calculation load is lower than a predetermined value, and determines the control amount of the plant 100 using the calculation result of a past nonlinear calculation when the calculation load is higher than a predetermined value. As a result, the optimization calculation method according to the embodiment can control the plant 100 using the nonlinear calculation result even when the calculation load is high, thereby improving the control accuracy. [Explanation of symbols]

[0022] 100 Plants

Claims

[Claim 1] When the calculation load is lower than a predetermined value, a control amount of the plant is determined in real time using a nonlinear calculation; When the calculation load is higher than the predetermined value, a control amount of the plant is determined by utilizing a result of a nonlinear calculation performed in the past. Optimization calculation method.

Images