Recent Advances and Applications of AI-Based Mathematical Modeling in Predictive Control of Hybrid Electric Vehicle Energy Management in China
Round 1
Reviewer 1 Report
Totally speaking, the topic is interesting, and the paper is well organized. However, I still provide the following comments, in order to improve the quality of the manuscript.
1. For the title, the first letter of each word should be capitalized, right? Please keep consistent.
2. Problem formulation is not clearly stated. Please give a clearer statement on the hybrid electric vehicle, MPC, and AI-based modeling.
3. As a review paper, the number of references is not enough and some of them are not tightly relevant, please cite more related references, credibility-based distributed frequency estimation for plug-in electric vehicles participating in load frequency control, resilient distributed frequency estimation for PEVs coordinating in load frequency regulation under cyber attacks.
4. From control methodologies point of view, Section 2.2 does not do a good job, the state-of-the-art control approaches are not thoroughly discussed.
5. “model predictive control” and “AI-based mathematical modeling” are not properly combined, as addressed in the title. More advanced-MPC related research is suggested to investigate, PSO-based model predictive control for load frequency regulation with wind turbines, intrusion-detector-dependent distributed economic model predictive control for load frequency regulation with PEVs under cyber attacks.
6. Some future works valuable to be done are suggested to be summarized. Authors can add a new section to give a summary on the promising future works from several aspects, based on the previous discussion.
7. There are some grammar mistakes and typos, please double check.
Author Response
Dear reviewers,
I deeply appreciate your advice and guidance on this paper, which has benefited me a lot. The following contents are the alterations based on reviewer’s comments, all of which have been highlighted in the article.
Totally speaking, the topic is interesting, and the paper is well organized. However, I still provide the following comments, in order to improve the quality of the manuscript.
- For the title, the first letter of each word should be capitalized, right? Please keep consistent.
Reply:I have revised it according to the suggestion.
- Problem formulation is not clearly stated. Please give a clearer statement on the hybrid electric vehicle, MPC, and AI-based modeling.
Reply:Hybrid electric vehicle is a kind of electric vehicle, which has two or more power devices, such as engine and motor. Therefore, hybrid electric vehicles combine the advantages of traditional fuel vehicles and pure electric vehicles, which can not only improve the fuel economy of vehicles, but also make up for the short range of pure electric vehicles. Hybrid electric vehicle model predictive control is to convert the global optimization problem of fuel economy under the whole driving condition into a local optimization problem in the finite time domain. The prediction model is used to obtain the vehicle demand torque, speed and other information in each prediction time domain. Through continuous rolling optimization, the model predictive control is applied to hybrid electric vehicles. The "energy management system" developed based on artificial intelligence algorithm is mainly applied to the management of energy and power distribution of plug-in hybrid electric vehicles, so as to achieve the purpose of energy conservation and emission reduction.
- As a review paper, the number of references is not enough and some of them are not tightly relevant, please cite more related references, credibility-based distributed frequency estimation for plug-in electric vehicles participating in load frequency control, resilient distributed frequency estimation for PEVs coordinating in load frequency regulation under cyber attacks.
Reply:I have updated the literature. Hu Z, Liu S, Wu L. Credibility-based distributed frequency estimation for plug-in electric vehicles participating in load frequency control[J]. International Journal of Electrical Power & Energy Systems, 2021, 130: 106997.
Hu Z, Liu J, Gao S, et al. Resilient Distributed Frequency Estimation for PEVs Coordinating in Load Frequency Regulation Under Cyber Attacks[C]//2021 IEEE 30th International Symposium on Industrial Electronics (ISIE). IEEE, 2021: 01-06.
- From control methodologies point of view, Section 2.2 does not do a good job, the state-of-the-art control approaches are not thoroughly discussed.
Reply:The goal of rule-based control strategy is to make the engine work at the optimal working point of efficiency, fuel economy and emissions at a specific engine speed. The control strategy of Toyota Prius and Honda Insight hybrid power system is based on this strategy. However, the engine power tracking control strategy has an obvious disadvantage that the overall efficiency of the powertrain is not optimal. The method based on fuzzy rules can realize the real-time and sub optimal energy distribution of the hybrid power system. Tate first applied the optimization algorithm tool to solve the energy management problem, and converted the fuel consumption optimization problem into a nonlinear convex optimization problem to obtain the global optimal solution. However, because the process of transforming nonlinear model into linear model is extremely complex, and linear programming is generally not applicable to complex nonlinear systems, the linear programming method is not very practical. In view of the fact that global optimization algorithms cannot be directly used for real-time control, the research on instantaneous optimization strategies that can be used for real-time control has become a hot topic. Sciarretta A has designed an Equivalent Consumption Minimization Strategy (ECMS) to achieve real-time optimization. By introducing the concept of equivalent factor to weight electric energy consumption, the sum of weighted electric energy consumption and fuel consumption is taken as the optimization goal. The disadvantage of this method is that the change range of battery state of charge is not clearly set, and the value of equivalent factor is only applicable to specific working conditions. Model Predictive Control (MPC) strategy can just make up for the large amount of calculation of global optimization algorithm and the inability of instantaneous optimization algorithm to achieve global optimization. Borhan H believes that the future demand torque of the vehicle attenuates in the form of an exponential function, and establishes a linear optimal control model. Compared with the rule-based control strategy, MPC achieves better fuel economy.
- “model predictive control” and “AI-based mathematical modeling” are not properly combined, as addressed in the title. More advanced-MPC related research is suggested to investigate, PSO-based model predictive control for load frequency regulation with wind turbines, intrusion-detector-dependent distributed economic model predictive control for load frequency regulation with PEVs under cyber attacks.
Reply:The idea of model predictive control is often combined with traffic information to establish a traffic information prediction model to make up for the lack of unknown future road information in practice, and combines with other optimization algorithms to calculate energy management problems. Some literatures take the hybrid vehicle fleet as the research object, and propose a MPC energy management method combining the road slope information. Some literatures have proposed an energy management method based on nonlinear MPC, which uses DP algorithm to solve the optimization problem in the prediction area by rolling to calculate specific energy management problems. This method shows its advantages in terms of fuel economy.
- Some future works valuable to be done are suggested to be summarized. Authors can add a new section to give a summary on the promising future works from several aspects, based on the previous discussion.
Reply:I added the prospect part to the conclusion. Considering the challenges faced by HEV energy management at present, the future development direction of HEV energy management strategy mainly focuses on the following aspects.
First, further optimize the control objectives, such as improving fuel economy, reducing emissions, and ensuring driveability. Second, further extraction and utilization of traffic information. Although it is easy to extract traffic information with the development of GPS, ITS and other technologies, due to the complexity and nonlinearity of traffic information, how to select targeted information and reasonable modeling is still a research hotspot. Third, accurate prediction of future information. At present, some scholars have proposed a variety of prediction methods and made great progress. However, in order to obtain the optimal energy distribution scheme, more accurate future information prediction methods still need to be developed.
- There are some grammar mistakes and typos, please double check.
Reply:Grammar mistakes and typos have been modified.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Avoid mass citation, especially in the introduction like [1-4]. Explain each reference in detail.
Are there only 28 references related to AI and EV energy management? The paper does not seem to have covered all the literature on this topic.
What are the references for the classification of control strategies? Each part should be cited to their main references.
The authors should elaborate on the obtained results.
What policies can be suggested by the results of this paper?
The conclusion highlights China, but there is nothing mentioned for China in the title.
Author Response
Dear reviewers,
I deeply appreciate your advice and guidance on this paper, which has benefited me a lot. The following contents are the alterations based on reviewer’s comments, all of which have been highlighted in the article.
Avoid mass citation, especially in the introduction like [1-4]. Explain each reference in detail.
Reply：I have modified the citation according to the suggestions.
Are there only 28 references related to AI and EV energy management? The paper does not seem to have covered all the literature on this topic.
Reply:According to the suggestions, I have added references related to artificial intelligence and electric vehicle energy management.
What are the references for the classification of control strategies? Each part should be cited to their main references.
Reply:There are usually two kinds of control objectives: one is to optimize the fuel consumption and emissions of the whole vehicle, and the other is to minimize the battery power consumption. Among them, the control strategy with the former as the control goal is the development direction of the energy management control strategy of hybrid electric vehicles. At the same time, in the process of setting the control goal, appropriate methods are taken to control the battery state of charge, which can give full play to the advantages of hybrid electric vehicles under the condition of ensuring the battery life.
The authors should elaborate on the obtained results.
Reply:This paper comprehensively summarizes the development of energy management technology of hybrid electric vehicles, summarizes the characteristics and defects of energy management strategies based on rules and optimization algorithms, analyzes the development direction of energy management strategies in recent years, and focuses on the energy management strategies of hybrid electric vehicles based on traffic information extraction and prediction.
What policies can be suggested by the results of this paper?
Reply:Due to the variability of HEV state, the accuracy of prediction model is the key to MPC control strategy. Therefore, we need to establish a more accurate vehicle prediction model, combine it with the intelligent transportation system, and use advanced sensors, GPS, and intelligent transportation system to make the vehicle prediction more closely linked with the actual working conditions. At the same time, most of the weighting factors, control time domain and prediction time domain values obtained by experience or experiment also play a key role in the whole strategy. It can be further optimized and analyzed or the method of condition identification can be introduced to obtain the most suitable value for the current state.
The conclusion highlights China, but there is nothing mentioned for China in the title.
Reply:China has been emphasized in the title.
Author Response File: Author Response.pdf
Reviewer 3 Report
Review report:
Authors reported “Recent advances and applications of AI-based mathematical modeling in Predictive Control of Hybrid Electric Vehicle Energy Management”. The organization of this work is good, and the discussion is well organized. Nevertheless, I have some comments which are listed below.
1. The Artificial intelligence explanation is not clear, and it should be revised in a detailed way.
2. The author claimed that their work is a novel investigation, however, myriads of works related to “Hybrid electric vehicle“ have been published to date. So, authors should change the way of the presentation focusing on novelty. The introduction should be improved with a paragraph describing the novelty and importance of the work.
3. The authors must carefully claim their novelty in the INTRODUCTION. In addition, the authors need to do some formatting errors that should be carefully checked and corrected in the text.
4. Please provide the comparison table, which you need to prove that your material is superior to previously reported literature.
5. Authors should be trimmed/condensed the ‘Abstract’ and ‘Conclusion’ sections in the revised manuscript. Please keep highlights of the whole manuscript in both sections.
Author Response
Dear reviewers,
I deeply appreciate your advice and guidance on this paper, which has benefited me a lot. The following contents are the alterations based on reviewer’s comments, all of which have been highlighted in the article.
Avoid mass citation, especially in the introduction like [1-4]. Explain each reference in detail.
Reply：I have modified the citation according to the suggestions.
Are there only 28 references related to AI and EV energy management? The paper does not seem to have covered all the literature on this topic.
Reply:According to the suggestions, I have added references related to artificial intelligence and electric vehicle energy management.
What are the references for the classification of control strategies? Each part should be cited to their main references.
Reply:There are usually two kinds of control objectives: one is to optimize the fuel consumption and emissions of the whole vehicle, and the other is to minimize the battery power consumption. Among them, the control strategy with the former as the control goal is the development direction of the energy management control strategy of hybrid electric vehicles. At the same time, in the process of setting the control goal, appropriate methods are taken to control the battery state of charge, which can give full play to the advantages of hybrid electric vehicles under the condition of ensuring the battery life.
The authors should elaborate on the obtained results.
Reply:This paper comprehensively summarizes the development of energy management technology of hybrid electric vehicles, summarizes the characteristics and defects of energy management strategies based on rules and optimization algorithms, analyzes the development direction of energy management strategies in recent years, and focuses on the energy management strategies of hybrid electric vehicles based on traffic information extraction and prediction.
What policies can be suggested by the results of this paper?
Reply:Due to the variability of HEV state, the accuracy of prediction model is the key to MPC control strategy. Therefore, we need to establish a more accurate vehicle prediction model, combine it with the intelligent transportation system, and use advanced sensors, GPS, and intelligent transportation system to make the vehicle prediction more closely linked with the actual working conditions. At the same time, most of the weighting factors, control time domain and prediction time domain values obtained by experience or experiment also play a key role in the whole strategy. It can be further optimized and analyzed or the method of condition identification can be introduced to obtain the most suitable value for the current state.
The conclusion highlights China, but there is nothing mentioned for China in the title.
Reply:China has been emphasized in the title.
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
No more comments.
Reviewer 3 Report
It can be accepted in its current format
