Engine calibration parameter optimization method based on automobile exhaust emission simulation environment

Abstract

The invention discloses an engine calibration parameter optimization method based on an automobile exhaust emission simulation environment, and the method comprises the steps: (1) constructing an automobile emission test simulation environment based on machine learning, and (2) searching an engine calibration optimal parameter based on the simulation environment. Low-cost and high-efficiency optimization of engine calibration parameters of automobile exhaust emission in a simulation environment is realized. According to the method, the automobile engine parameters are optimized efficiently atlow cost in the simulation environment, and the technical effects of cost reduction and efficiency increase are achieved.

Description

technical field [0001] The invention relates to a method for optimizing engine calibration parameters based on simulation training. In a simulation environment based on machine learning technology, the vehicle exhaust emission process is simulated to obtain an optimization strategy for engine calibration parameters, which belongs to the field of industrial intelligent manufacturing. Background technique [0002] The vehicle engine calibration process adjusts the parameters of each control module in the electronic control unit to make the exhaust gas emitted during driving meet the national emission standards. After the engine, control strategy, and peripheral devices are determined, in order to make the vehicle's emissions meet the national standards, it is necessary to find out the parameters of the engine under working conditions. The general calibration process is first to make the engine work normally through the bench test, and then get the initial set of parameters. N...

AI Technical Summary

Problems solved by technology

The cost of a complete vehicle test is as high as tens of thousands, while parameter calibration is a sequential decision-making process of iterative optimization. To complete the calibration of an engine often requires hundreds of vehicle tests, which is costly and unavoidable

[0006] (2) Highly dependent on the experience of the calibration personnel

However, in the process of reinforcement learning, the process of a large amount of interactive trial and error between the agent and the environment is often difficult to implement in the real environment.

For example, if reinforcement learning is used directly in the actual car testing process, it is necessary to use a large number of different parameter configurations to conduct trial and error in the real bench test, and then learn the optimal model from the car interaction trajectory data under these parameter conditions. parameters, obviously, this method will bring huge economic loss and time cost, which is unacceptable

Images