Combined Optimization Research Articles

Evaluating the Effect of the Communication Link of the Relays on the Operation Time of the Protection System

In case of an error in the power system, it is the duty of the protection relays to isolate the faulty part from the rest of the system as soon as possible. Overcurrent and distance relays are usually used to protect the transmission and super distribution systems. The optimum coordination of these relays is very important. In this article, communication links are used to improve the optimal coordination of overcurrent and distance relays. In the proposed plan, the overcurrent relays on both sides of the line are equipped with a communication link for immediate action in the event of an error. Using this communication link, the complete coverage of the line is also provided by distance relays. The number and place of installation of the communication link greatly affect the operation time of overcurrent relays and the distance and coordination between them. The coordination problem is formulated by considering the location and number of communication links, and to solve it, the combined optimization algorithm of genetics and linear programming will be used. With the help of the genetic algorithm, the location and setting of the second zone of the distance relay, and by using the linear programming algorithm, the setting of the overcurrent relays is determined. The proposed method has been tested on the IEEE 14-bus (the simulation in MATLAB software) network and its results examined, and it will be seen how much the existence of a communication link can improve the performance time of the protection system.

A Ridepooling Bidding Model Considering Service Quality from the Perspective of Cournot Game Theory

Pricing for ridepooling services may be affected by travel needs, penalties, and ride service quality. Firstly, this paper explores pricing methods based on the perceived value of the ride as compared to the distance–time prices and penalties and proposes a ridepooling travel utility function for pricing strategies. Then this paper develops a combined optimization ridepooling scheduling model integrating order allocation and path optimization (CORM), and designs a hybrid simulated annealing (HSA) heuristic algorithm for solving this model. Based on the Cournot game theory, this paper further develops a ridepooling bidding model (RBM) that can simultaneously optimize the seekers’ utility and the profit of the ridepool platform. The effectiveness of the CORM and HSA are verified, and the superiority of the RBM has also been proved by a numerical simulation example. The results show the following: (1) neither the low distance–time price and high penalty price strategy nor the high distance–time and low penalty price strategy is an absolute advantage strategy for ridepooling passengers and the platform; (2) compared with the CORM, the RBM can not only effectively improve the profit of ridepooling platform, but also improve the success rate of ridepooling.

Trispecific antibodies for cancer immunotherapy.

Despite the clinical success of monoclonal and bispecific antibodies, there are still limitations in the therapeutic effect of malignant tumours, such as low response rate, treatment resistance, and so on, inspiring the exploration of trispecific antibodies (TsAbs). TsAbs further improve the safety and efficacy and has great clinical potential through three targets combination and formats optimization. This article reviews the development history and the target combination features of TsAbs. Although there are still great challenges in the clinical application of TsAbs, it is undeniable that TsAbs may be a breakthrough in the development of antibody drugs.

On the multi-period combined maintenance and routing optimisation problem

This paper focuses on the combined maintenance and routing optimisation problem in a multi-period environment that consists of scheduling maintenance operations for a set of geographically distributed machines, subject to non-deterministic failures with a set of technicians that perform preventive maintenance and corrective operations. This study uses a two-step approach based on column generation. The first step consists of a maintenance model that determines the optimal time until the next preventive maintenance operation, and each machine’s maintenance frequency while minimising the total expected maintenance costs. The second step involves a routing model that assigns and schedules maintenance operations for each technician over the planning horizon while minimising the routing and maintenance costs. We formulate the second step problem as a periodic vehicle routing problem and to solve it, we propose a column generation approach. The master problem is a set partitioning formulation that indicates which machines must be visited in the planning horizon. This decomposition has two auxiliary problems: the first one is an elementary shortest-path problem to find a feasible path for each period and the second one finds a set of visiting periods for each machine. Our approach balances the maintenance cost, routing cost, and failure probabilities.

Computational design of clinical trials using a combination of simulation and the genetic algorithm.

Artificial intelligence (AI) has come to be used in various technological fields in recent years. However, there have been no reports of AI-designed clinical trials. In this study, we tried to develop study designs by a genetic algorithm (GA), which is an AI solution for combination optimization problems. Specifically, the computational design approach was applied to optimize the blood sampling schedule for a bioequivalence (BE) study in pediatrics and optimize the allocation of dose groups for a dose-finding study. The GA could reduce the number of blood collection points from 15 (typical standard) to seven points without meaningful impact on the accuracy and precision of the pharmacokinetic estimation for the pediatric BE study. For the dose-finding study, up to 10% reduction of the total number of required subjects from the standard design could be achieved. The GA also created a design that would lead to a drastic reduction of the required number of subjects in the placebo arm while keeping the total number of subjects at a minimum level. These results indicated the potential usefulness of the computational clinical study design approach for innovative drug development.

Evaluation of fatigue cracks on rail material based on laser nonlinear wave modulation and adaptive particle swarm optimization - support vector machines

ABSTRACT The evaluation of fatigue cracks is essential for the detection of early damage to railway tracks. Laser nonlinear wave modulation spectroscopy using broadband excitation is an effective method for fatigue crack detection that can solve the problem of frequency combination optimisation. However, due to the complex components of laser nonlinear ultrasonic signals, the evaluation of fatigue cracks using laser nonlinear ultrasonic signals remains a challenge. In this study, wavelet packet decomposition and principal component analysis were used to extract various features of nonlinear ultrasonic signals in the frequency domain, and four new combinations of features were obtained. These new features were used as input for the evaluation model. Finally, a fatigue crack evaluation model based on adaptive particle swarm optimisation-support vector machines was proposed to accurately evaluate fatigue cracks of different lengths. The experiments conducted showed that the fourth new combination of features and evaluation model proposed in this paper can improve the accuracy of evaluating rail fatigue cracks of different lengths, reaching a classification accuracy of 97%.

On the Effects of Optimal Implementation of Variable Rotor Speed and Power Management on Hybrid-Electric Rotorcraft

AbstractThe concept of variable rotor speed (VRS) has been recognized as an efficient means to improve rotorcraft operational performance and environmental impact, with electrification being a potential technology to further contribute to that. This paper explores the impact of optimal implementation and scheduling of VRS and power management strategy for conventional and hybrid-electric rotorcraft on energy, fuel, and emissions footprint. A multidisciplinary simulation framework for rotorcraft performance combined with models for engine performance and gaseous emissions estimation is deployed. A holistic optimization approach is developed for the derivation of globally optimal schedules for combined rotor speed and power split targeting minimum energy consumption. Application of the derived optimal schedules at mission level resulted to a 6% improvement in range capability for the VRS tilt-rotor relative to its conventional counterpart. For the hybrid-electric tilt-rotor, combined optimization of VRS and power management leads to an increase in range to 18.4% at 40% and 25% reduced payload for current (250 Wh/kg) and future (450 Wh/kg) battery technology, respectively. For representative urban air mobility (UAM) scenarios, it is demonstrated that the VRS concept resulted in up to 10% and 14% reductions in fuel burn and NOx relative to the nominal rotor speed case, respectively. The utilization of the combined optimum VRS and power split schedules can boost performance with reductions of the order of 20% and 25% in mission fuel/CO2 and NOx at a reduced payload relative to the conventional tilt-rotor.

Microservice combination optimisation based on improved gray wolf algorithm

Microservices architecture is a new paradigm for application development. The problem of optimising the performance of microservice architectures from a non-functional perspective is a typical Nondeterministic Polynomial (NP) problem. Therefore, aiming to quantify the non-functional requirements of computing microservice systems, while solving the problem of latency in computing the best combination of services with the maximum QoS objective function value, this paper proposes a microservice combination approach based on the QoS model and a CGWO algorithm for optimisation computation for this model. The experimental results verify that the error rate of the method is only 0.528% on the non-functional combination optimisation problem, and the computational efficiency of the algorithm increases by 97.29% when the complexity of the problem search space increases, while CGWO improves 65.97% and 81.25% respectively in the accuracy of optimisation compared to the prototype of the algorithm (GWO), and has a stable optimisation performance, aspect. It proves that the research in this paper has a high advantage in automatically searching for the best QoS for the microservice combination problem.

Stability metering method for high-resolution space telescope with large F-number based on combination optimization

The development of a lightweight and compact high-resolution space telescope with a large F-number is urgently required for rapid networking of the remote sensing smallsat constellation. Considering the limitation of high-resolution imaging and optical performance, a stability metering method is investigated based on combination optimization in this paper, it follows that a novel integrated metering structure composed of carbon fiber reinforced polymer, with high stiffness, athermalization, and a mass fraction of 15 % in a space telescope, is presented. According to the error function of secondary mirror misalignment and the ABD matrix of laminate, the performance of the integrated metering structure is improved gradually via combination optimization. The finite element and optical analyzes demonstrated that the secondary misalignments satisfy the optical tolerance and the modulation transfer function attenuation ratio is only 1.85 %, moreover, performance tests of a prototype are performed and the results are in good agreement with the optimization and analysis. These results prove the feasibility and validity of the proposed method. This study solves the problem of stable metering for large F-number optical systems and provides a basis for the development of space telescopes.

Combination mode and optimization strategy of harvest procedure of anterolateral thigh chimeric perforator myocutaneous flap

To summarize the combination methods and optimization strategies of the harvest procedure of anterolateral thigh chimeric perforator myocutaneous flap. A clinical data of 359 cases of oral cancer admitted between June 2015 and December 2021 was retrospectively analyzed. There were 338 males and 21 females with an average age of 35.7 years (range, 28-59 years). There were 161 cases of tongue cancer, 132 cases of gingival cancer, and 66 cases of buccal and oral cancer. According to the Union International Center of Cancer (UICC) TNM staging, there were 137 cases of T 4N 0M 0, 166 cases of T 4N 1M 0, 43 cases of T 3N 1M 0, 13 cases of T 3N 2M 0. The disease duration was 1-12 months (mean, 6.3 months). The soft tissue defects in size of 5.0 cm×4.0 cm to 10.0 cm×7.5 cm remained after radical resection were repaired with the free anterolateral thigh chimeric perforator myocutaneous flaps. The process of harvesting the myocutaneous flap was mainly divided into 4 steps. Step 1: exposing and separating the perforator vessels, which mainly came from the oblique branch and the lateral branch of the descending branch. Step 2: isolating the main trunk of the perforator vessel pedicle and determining the origin of the vascular pedicle of muscle flap, which was came from oblique branch, lateral branch of the descending branch, or medial branch of the descending branch. Step 3: determining the source of muscle flap, including lateral thigh muscle and rectus femoris muscle. Step 4: determining the harvest form of muscle flap, which included muscle branch type, main trunk distal type, and main trunk lateral type. The 359 free anterolateral thigh chimeric perforator myocutaneous flaps were harvested. In all cases, the anterolateral femoral perforator vessels existed. The perforator vascular pedicle of the flap came from the oblique branch in 127 cases and the lateral branch of the descending branch in 232 cases. The vascular pedicle of muscle flap originated from the oblique branch in 94 cases, the lateral branch of the descending branch in 187 cases, and the medial branch of the descending branch in 78 cases. The muscle flaps harvested from the lateral thigh muscle in 308 cases and the rectus femoris muscle in 51 cases. The harvest forms of muscle flaps included 154 cases of muscle branch type, 78 cases of main trunk distal type, and 127 cases of main trunk lateral type. The size of skin flaps ranged from 6.0 cm×4.0 cm to 16.0 cm×8.0 cm, and the size of muscle flaps range from 5.0 cm×4.0 cm to 9.0 cm×6.0 cm. In 316 cases, the perforating artery anastomosed with the superior thyroid artery, and the accompanying vein anastomosed with the superior thyroid vein. In 43 cases, the perforating artery anastomosed with the facial artery, and the accompanying vein anastomosed with the facial vein. After operation, the hematoma occurred in 6 cases and vascular crisis in 4 cases. Among them, 7 cases were successfully saved after emergency exploration, 1 case had partial necrosis of skin flap, which was healed after conservative dressing change, and 2 cases had complete necrosis of skin flap, which was repaired by pectoralis major myocutaneous flap. All patients were followed up 10-56 months (mean, 22.5 months). The appearance of the flap was satisfactory, and the swallowing and language functions were restored satisfactorily. Only linear scar left in the donor site with no significant effect on thigh function. During follow-up, 23 patients had local tumor recurrence and 16 patients had cervical lymph node metastasis. The 3-year survival rate was 38.2% (137/359). The flexible and clear classification of the key points in the harvest process of anterolateral thigh chimeric perforator myocutaneous flap can optimize the protocol to the greatest extent, increase the safety of the operation, and reduce the difficulty of the operation.

Multi-scale collaborative optimization of SrTiO3-based energy storage ceramics with high performance and excellent stability

SrTiO3 (ST)-based ceramics are considered as promising candidates for energy storage applications. However, the low polarization intensity in ST-based materials limits their energy storage performance, rendering materials that usually exhibit a low recoverable energy-storage density. In the present study, we have optimized the energy storage performance of ST-based ceramics by using a combined optimization strategy of structural engineering and microstructural regulation. High performance (Sr1-x-y-2φNayBixCaφ□φ)TiO3 (abbreviated as zSNBCT, where □ represents the Sr vacancies) ceramics were thereby designed. During composition optimization, the phase state of zSNBCT was adjusted to a critical point where the relaxor ferroelectric-paraelectric phase transition occurred around room temperature. It thus induced a strong relaxation behavior with the formation of ferroelectric polar nano-regions, yielding a high recoverable energy-storage density (Wrec) of ∼6 J/cm3 and a high energy-storage efficiency (η) of ∼92% under a large breakdown electric field of 440 kV/cm, for z = 0.2 sample. Moreover, the breakdown strength (BDS) of the 0.2SNBCT ceramic was further improved by adopting a two-step sintering and spark plasma sintering approach for the microstructural refinement. Simulation models containing grains and grain boundaries were well established using phase-field simulation and finite element analysis. These simulations came to a similar conclusion for the enhanced BDS. Namely, the fine-grained microstructure significantly hindered the growth of breakdown cracks under an applied electric field. Most importantly, the 0.2SNBCT sample showed excellent frequency stability (1−1000 Hz), thermal stability (20−140 °C), and cycling stability (105 cycles), rendering it a promising candidate for energy storage systems. Our designed strategy of structural engineering and microstructural regulation may provide a new paradigm for the design of high-performance energy storage ceramics for pulse power applications.

Rapid evaluation of Radix Paeoniae Alba and its processed products by near-infrared spectroscopy combined with multivariate algorithms.

It is well known that the processing method of herbal medicine has a complex impact on the active components and clinical efficacy, which is difficult to measure. As a representative herb medicine with diverse processing methods, Radix Paeoniae Alba (RPA) and its processed products differ greatly in clinical efficacy. However, in some cases, different processed products are confused for use in clinical practice. Therefore, it is necessary to strictly control the quality of RPA and its processed products. Giving that the time-consuming and laborious operation of traditional quality control methods, a comprehensive strategy of near-infrared (NIR) spectroscopy combined with multivariate algorithms was proposed. This strategy has the advantages of being rapid and non-destructive, not only qualitatively distinguishing RPA and various processed products but also enabling quantitative prediction of five bioactive components. Qualitatively, the subspace clustering algorithm successfully differentiated RPA and three processed products, with an accuracy rate of 97.1%; quantitatively, interval combination optimization (ICO), competitive adaptive reweighted sampling (CARS), and competitive adaptive reweighted sampling combined with successive projections algorithm (CARS-SPA) were used to optimize the PLS model, and satisfactory results were obtained in terms of wavelength selection. In conclusion, it is feasible to use NIR spectroscopy to rapidly evaluate the effect of processing methods on the quality of RPA, which provides a meaningful reference for quality control of other herbal medicines with numerous processing methods.

Isogeometric vibration and material optimization of rotating in-plane functionally graded thin-shell blades with variable thickness

A combined formulation is proposed for modeling and material optimization of rotating in-plane functionally graded (IFG) thin-shell blades with variable thickness in this paper. The aim of this work is to study the vibration characteristics and optimize the IFG volume fraction of the rotating blade model. The isogeometric approach (IGA) and metaheuristic algorithms are employed to find the compositional profile of a two-constituent material blade with the goal of maximizing the fundamental frequency. By considering the rotating speed, the vibrational behaviors of the blades are predicted based on the thin-shell theory and IGA. The different effects due to the rotating speed, variable thickness and IFG materials are considered in the governing equations. For the optimization model, the high-order non-uniform rational B-spline (NURBS) can ensure a smooth material composition profile with fewer design variables. The changing of material composition at control points in plane provides more possibilities for blade material design. Several comparison studies are performed to verify the accuracy of current models for dealing with the vibration behaviors and optimizing the material distribution of the thin-shell blades. The correctness of the combined optimization approach is also proved by comparing those optimal results of two algorithms. Through the optimization process, the optimal material distribution of thin-shell blades with uniform and variable thickness is obtained.

Investigation of image-based lesion and kidney dosimetry protocols for 177Lu-PSMA-I&T therapy with and without a late SPECT/CT acquisition

Background177Lu-PSMA therapy has been successfully used to prolong the survival of patients with metastatic castration-resistant prostate cancer. Patient-specific dosimetry based on serial quantitative SPECT/CT imaging can support the understanding of dose–effect relationships. However, multiple SPECT/CT measurements can be challenging for patients, which motivates the investigation of efficient sampling schedules and their impact on dosimetry. In this study, different time samplings with respect to the number and timing of SPECT/CT acquisitions with and without a late measurement were investigated.Materials and methodsIn total, 43 lesions and 10 kidneys of 5 patients receiving 177Lu-PSMA-I&T therapy were investigated. Whole-body SPECT/CT measurements were performed at 1, 2, 3 and 7 days post-injection. For both lesions (isocontour-based segmentation) and kidneys (CT-based segmentation), a reference model was employed including all four time points. To identify the best-matching fit function out of a pre-defined set of models, visual inspection, coefficients of variation and sum of squared errors were considered as goodness-of-fit criteria. Biologically effective doses (BEDs) calculated with different time samplings (days 1, 2, 3/1, 2, 7/1, 3, 7/2, 3, 7 and 1, 2/1, 3/1, 7) were compared to the reference.ResultsThe best-fit function was found to be a mono-exponential model for lesions and a bi-exponential model with a population-based parameter and two free parameters for kidneys. The BEDs calculated with the time sampling 1, 3, 7 days showed the lowest deviations from the reference for lesions with 4 ± 5%. Without day 7, still 86% of all lesions showed deviations from the reference < 10%. The outlier deviations showed a positive correlation with the effective half-life of the respective lesions. For kidneys, including days 1, 2, 3 achieved the best results with 0 ± 1%. Generally, deviations for kidneys were found to be small for all time samplings (max. 13%).ConclusionsFor combined optimization of the SPECT/CT time sampling for kidney and lesion dosimetry during 177Lu-PSMA-I&T therapy, the sampling with days 1, 3, 7 showed the smallest deviation from the reference. Without a late acquisition, using the schedule with days 1, 2, 3 is likewise feasible.

Multi-granularity source-load-storage cooperative dispatch based on combined robust optimization and stochastic optimization for a highway service area micro-energy grid

Integrating renewable energy into planning and operation of transportation infrastructures can help to promote the various sector collaborative decarbonization. For the highway service area micro-energy grid (HSAMEG), its optimization lacks the source-load-storage cooperation and the modeling that considers both accuracy and complexity, and is hard to balance reliability and flexibility due to uncertainties in renewable energy and charging-demand. For these issues, a novel dispatch is proposed to balance the dispatch reliability and flexibility, and the model accuracy and complexity by combining advantages of robust and stochastic optimizations and applying the multi-granularity modeling. First, the source-load-storage configuration is established. Then the multi-granularity model is developed by fine-grained model based on the operating characteristics and coarse-grained model based on the equivalent energy storage characteristics. Finally, based on distribution characteristics of online-optimization forecast-errors, a multi-granularity source-load-storage cooperative dispatch combining robust optimization and stochastic optimization is proposed. The simulation results show that the source-load-storage collaboration increases the self-contained objective by 10%. Compared with robust optimization, the proposed strategy enhances the economic objective by 17% and the self-contained objective by 16.2%. Compared with stochastic optimization, the proposed strategy improves the computation efficiency by over 5 times and the self-contained objective by 8.8% without constraint violations.

A New Hybrid Grasshopper Optimization - Backpropagation for Feedforward Neural Network Training

The Grasshopper optimization algorithm showed a rapid converge in the initial phases of the global search, however while being around the global optimum, the searching process became so slow. On the contrary, the gradient descending method around achieved faster convergent speed global optimum, and the convergent accuracy was showed to be higher at the same time. As a result, the proposed hybrid algorithm combined Grasshopper optimization algorithm (GOA) along with the back-propagation (BP) algorithm, also referred to as GOA–BP algorithm, was introduced to provide training to the weights of the feed forward neural network (FNN), the proposed hybrid algorithm can utilize the strong global searching ability of the GOA, and the intense local searching ability of the Back-Propagation algorithm. The results of experiments showed that the proposed hybrid GOA–BP algorithm was better and faster in convergent speed and accuracy than the Grasshopper optimization algorithm (GOA) and BP algorithm.

Development of a Novel Low-Calorie Lime Juice-Based Prebiotic Beverage Using a Combined Design Optimization Methodology.

A novel lime-juice based low-calorie functional beverage was developed by using D-optimal combined design optimization. For the preparation of the beverage, the following functional ingredients were used: lime juice, lime peel essential oil (LEO) as a flavoring agent and bioactive component, sucralose as a low-calorie sweetener, an inulin/polydextrose (I/P) mixture as prebiotic fibers, pectin as a thickening agent and soluble dietary fiber, lutein as a carotenoid colorant and antioxidant, and peppermint extract (ME) as a flavoring agent and bioactive component. A combined design consisting of one mixture factor (LEO/ME ratio), one numeric factor (lutein concentration), and one categoric factor (presence or absence of prebiotics) was used for optimizing the functional beverage based on the sensory quality. Regression models were adequately fitted to the data of sensory acceptance with a determination coefficient >90%. The sample containing a mixture of prebiotics, 2:3 (v/v) ratio of LEO: ME, and 3 mg/100 mL lutein was selected as the best formulation among the six optimal beverages which was suggested by Design-Expert software. This final optimum sample showed the highest total phenolic (44.22 mg gallic acid equivalents/L) and flavonoid (25.49 mg quercetin equivalents/L) contents, and its antioxidant activity (as 2,2-diphenyl-1-picrylhydrazyl radical (DPPH•) scavenging) was 38.30%. The newly designed beverage has the potential to promote health benefits and in therapeutic applications.

Lipoprotein detection: Hybrid deep classification model with improved feature set

Patients with chronic liver diseases typically experience lipid profile problems, and mortality from cirrhosis complicated by portal vein thrombosis (PVT) is very significant. A lipoprotein (Lp) is a bio-chemical assemblage with the main job of moving fat molecules in water that are hydrophobic. Lipoproteins are present in all eubacterial walls. Lipoproteins are of tremendous interest in the study of spirochaetes’ pathogenic mechanisms. Since spirochaete lipobox sequences are more malleable than other bacteria, it’s proven difficult to apply current prediction methods to new sequence data. The major goal is to present a Lipoprotein detection model in which correlation features, enhanced log energy entropy, raw features, and semantic similarity features are extracted. These extracted characteristics are put through a hybrid model that combines a Gated Recurrent Unit (GRU) and a Long Short-Term Memory (LSTM). Then, the outputs of GRU and LSTM are averaged to obtain the output. Here, GRU weights are optimized via the Selfish combined Henry Gas Solubility Optimization with cubic map initialization (SHGSO) model.

On genetic algorithm and artificial neural network combined optimization for a Mars rotorcraft blade

The Mars rotorcraft can help the Mars rover to optimize the driving route and explore more efficiently. However, the thin and cold Martian atmosphere can heavily increase the consumed power of the rotorcraft blades and reduce the generated thrust, leading to the reduction in power loading. The aerodynamic force generated by the Mars rotorcraft blade can be significantly influenced by the blade structure, which has several main structural characteristics and needs to be well designed. A multivariable optimization design method of blade structure is proposed based on neural network, genetic algorithm, and computational fluid dynamics simulations. Experiments conducted in the Martian atmosphere simulator indicate that the Mars rotorcraft blades with the hyperbolic chord length distribution and twist have higher power loading in the constant collective pitch angle. • A model was built to predict rotor hover performance at low-Reynolds-number. • Established an optimization model of the planform of the Mars rotorcraft blade. • The optimized blade consumes 29% less power than the conventional blade. • The optimization rusult were verified experimentally in Mars simulated environment.

PeerRemove: An adaptive node removal strategy for P2P botnet based on deep reinforcement learning

Botnets have become one of the major intrusion threats to cybersecurity. P2P botnets have high concealment and resilience because of their distributed structure, which are difficult to be completely dismantled and destroyed. Existing methods based on traffic statistics and vulnerability tracing cannot effectively solve the problem of P2P botnet disintegration. Although P2P networks emphasize that each node is peer-to-peer, the difference in processing power, resource distribution, and node bandwidth can lead to a certain heterogeneity. The critical nodes bridge the underlying bot nodes and the upper control server. Traditional methods for ranking the importance of nodes mainly relies on classical graph-theoretic feature statistics method, such as degree, betweenness, clustering coefficient, feature vector centrality, PageRank, etc. In this paper, botnet defense strategies are investigated from the perspective of complex network graph theory, and graph embedding and deep reinforcement learning combination optimization methods are adopted to handle the critical nodes identification problem of P2P botnets. Then, a novel adaptive node removal model called PeerRemove is proposed. The model uses Structure2vec graph embedding to characterize the network structure information as a low-dimensional embedding space, and it uses n-step Q-learning to train the model to learn complex topological patterns to find the critical nodes that effectively disintegrate the network. To evaluate the effectiveness of the proposed method, the Area Under the Curve (AUC) of the Largest Connected Component (LCC) size during node removal is used as an evaluation indicator, and six different types real or synthetic P2P botnets are selected, namely Sality, ZeroAccess, NSIS, Mozi, Gnutella, and Peer sampling service. Experiments are conducted on many real and model networks with node sizes reaching thousands and tens of thousands, and our method is compared with five classical static or dynamic node attack methods of HAD, PageRank, CI, BPD, and HPRA. The experimental results show that the overall AUC curve of the PeerRemove method is lower than that of the benchmark method, which can minimize botnet resiliency at a small cost. The proposed method is superior to the existing node removal methods and shows good robustness and feasibility. To demonstrate the generality of this method, it is tested on a centralized topological dataset and good experimental results are obtained.