Engineering machinery industry competitiveness evaluation and optimization method based on multi-dimensional analysis

Abstract

The invention relates to the technical field of industrial analysis, and discloses an engineering machinery industry competitiveness evaluation and optimization method based on multi-dimensional analysis, and the method comprises the steps: fusing industrial machinery industry multi-source data, competition pattern path evolution, scene environment simulation, reinforcement learning optimization algorithm and other technical means; accurate description of the strategy income relation and dynamic prediction of the competition pattern of the industry subject in different market scenes are realized. Meanwhile, optimization action income simulation is carried out by utilizing a trained strategy network in a target virtual market environment, strategy decision parameters are iteratively adjusted on this basis, and an optimization action sequence with high adaptability is generated, so that not only is the accuracy of competitiveness evaluation improved, but also personalized optimization scheme generation under multi-target constraint is supported, and the competitiveness evaluation efficiency is improved. And the strain capacity and decision-making scientificity of the engineering machinery industry in a complex market environment are obviously enhanced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of industrial analysis, and in particular to an engineering machinery industry competitiveness evaluation and optimization method based on multi-dimensional analysis. BACKGROUND

[0002] In the field of engineering machinery industry, competitiveness evaluation and optimization are important foundations to support enterprise strategic adjustment and market expansion. However, the existing technology mainly relies on static economic indicators and traditional models to analyze the industrial situation, which is difficult to reflect the complex dynamic relationship between enterprises within the industrial system. Especially with the rapid changes in the industrial environment, enterprises often lack accurate prediction support in the process of resource scheduling, market selection and product structure adjustment, leading to decision delay or even failure. Especially when involving multiple market scenarios, the existing evaluation method cannot effectively handle multi-source heterogeneous data and the linkage characteristics of upstream and downstream of the industrial chain, resulting in evaluation lag, prediction distortion, and lack of flexibility in optimization strategies, which is difficult to adapt to changing market demand and complex competitive situation, which not only affects the efficiency of enterprise resource allocation, but also weakens the coordination ability of the entire industrial chain, and restricts the improvement of industry competitiveness. SUMMARY

[0003] Therefore, the present application aims to provide an engineering machinery industry competitiveness evaluation and optimization method based on multi-dimensional analysis to solve the technical problems that the competitiveness of the engineering machinery industry cannot be effectively evaluated and optimized in the prior art, resulting in low efficiency of enterprise resource allocation, weakened coordination ability of the industrial chain, and restricted industry competitiveness.

[0004] The present application discloses an engineering machinery industry competitiveness evaluation and optimization method based on multi-dimensional analysis, which comprises:

[0005] A competition pattern change model is constructed based on multi-source data of the engineering machinery industry, and the strategy benefit relationship of each industrial subject in different market scenarios is analyzed through the competition pattern change model to predict the change path of the competition pattern;

[0006] A target virtual market environment is constructed based on the predicted competition pattern change path through a scenario simulation environment model;

[0007] In the target virtual market environment, the benefit changes of different optimization actions in the multi-market scenario are simulated through a reinforcement learning optimization algorithm, the strategy decision parameters are adjusted multiple times according to the simulation results, and an optimization scheme is determined.

[0008] Further, after the optimization scheme is determined, the optimization scheme is executed, and the parameters of the competition pattern change model, the reinforcement learning optimization algorithm and the scenario simulation environment model are adjusted based on the execution effect of the optimization scheme.

[0009] Further, the process of constructing a competitive pattern change model based on multi-source data of the engineering machinery industry specifically comprises:

[0010] Based on the multi-source data of the engineering machinery industry, historical strategy behavior data, market transaction records, supply chain response data, and external macroeconomic indicators of each main body in the industry chain are analyzed as the first data analysis result;

[0011] According to the first data analysis result, a state transition matrix and a strategy benefit matrix are constructed; the state transition matrix is used to describe the change law between different market scenarios, and the strategy benefit matrix represents the strategy selection and benefit change relationship of each main body in different market scenarios;

[0012] Based on the state transition matrix and the strategy benefit matrix, the behavior probability distribution of the main body is updated, and the change path of the industry competition pattern in the future time period is determined according to the behavior probability distribution of the main body.

[0013] Further, the virtual market environment is used to simulate the dynamics of the industry chain, market demand changes, and competitive strategy behavior responses.

[0014] Further, the target virtual market environment is constructed based on the predicted competitive pattern change path through the scenario simulation environment model.

[0015] Based on the predicted competitive pattern change path, the dynamic parameters of the industry chain, the market demand fluctuation parameters, and the competitive strategy behavior response parameters are extracted;

[0016] The parameters are fused in the scenario simulation environment model to generate a virtual market environment;

[0017] The boundary parameters of the virtual market environment are adjusted through environmental constraint conditions to obtain a target virtual market environment.

[0018] Further, the boundary parameters of the virtual market environment are adjusted through environmental constraint conditions.

[0019] Based on market supply and demand balance constraints, industry chain resource utilization constraints, and competitive strategy game constraints, the boundary parameters are dynamically adjusted through a multi-objective optimization algorithm to minimize the deviation between the virtual market environment and the actual market state.

[0020] Further, before constructing the scenario simulation environment model, the method further comprises:

[0021] Based on the multi-source data of the engineering machinery industry, an environment factor set is constructed; the environment factor set includes market demand, industry chain resource constraints, and competitive strategy behavior;

[0022] According to the historical data and the competitive pattern change path prediction result, the weight proportion of each environment factor is adjusted.

[0023] Further, the construction process of the scenario simulation environment model comprises:

[0024] According to the environment factor set and the weight proportion, the transition relationship between different market scenarios is determined, and a scenario simulation environment model is constructed based on the environment factor set, the weight proportion and the transition relationship between different market scenarios.

[0025] Further, the simulation of the revenue change of different optimization actions in the multi-market scenario through the reinforcement learning optimization algorithm in the target virtual market environment specifically comprises:

[0026] Based on the target virtual market environment, a state space, an action space and a reward mechanism are constructed;

[0027] Under the constraint of the constructed space and reward mechanism, the revenue change of each optimization action in the multi-market scenario is simulated based on the trained policy network, and an action-revenue mapping relationship is constructed;

[0028] Based on the action-revenue mapping relationship, the policy selection logic in the reinforcement learning optimization algorithm is adjusted.

[0029] Further, the iterative adjustment of the policy decision parameters based on the simulation result and the determination of the optimization scheme specifically comprise:

[0030] Based on the adjusted policy selection logic, the policy parameters are optimized through multiple rounds of training operations;

[0031] According to the training result, a preliminary optimization action sequence is generated in the target virtual market environment;

[0032] After executing the preliminary optimization action sequence, the sequence is adjusted based on the simulation feedback, the revenue maximization constraint, the resource constraint and the market risk factor to generate a final optimization action sequence;

[0033] The final optimization action sequence is used to determine an optimization scheme comprising capacity allocation, product combination adjustment and market entry path.

[0034] Compared with the prior art, the beneficial effects of the present application are:

[0035] The application realizes accurate description of the strategy income relationship of the industry subject under different market scenarios and dynamic prediction of the competition pattern by fusing industrial machinery industry multi-source data, competition pattern path evolution, scene environment simulation and reinforcement learning optimization algorithm and other technical means. Meanwhile, by using the trained strategy network to simulate the income change of different optimization actions in the target virtual market environment, the strategy decision parameters are iteratively adjusted on the basis to generate adaptive optimization action sequences, which not only improves the accuracy of the competitiveness evaluation, but also supports the generation of individualized optimization schemes under multi-objective constraints, significantly enhancing the adaptability and decision scientificity of the engineering machinery industry in the complex market environment. BRIEF DESCRIPTION OF DRAWINGS

[0036] The drawings described herein are used to provide a further understanding of the embodiments of the application, constitute a part of the application, and do not constitute a limitation of the embodiments of the application. In the drawings:

[0037] Figure 1 A flowchart of a multi-dimensional analysis-based engineering machinery industry competitiveness evaluation and optimization method disclosed by the embodiments of the application. DETAILED DESCRIPTION

[0038] In order to enable personnel in the art to better understand the application scheme, the technical solutions in the embodiments of the application will be described clearly and completely below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only a part of the embodiments of the application, not all the embodiments.

[0039] Embodiment one

[0040] The first aspect of the application discloses a multi-dimensional analysis-based engineering machinery industry competitiveness evaluation and optimization method. Please refer to Figure 1 , Figure 1 A flowchart of a multi-dimensional analysis-based engineering machinery industry competitiveness evaluation and optimization method disclosed by the embodiments of the application, the method comprising:

[0041] A competition pattern change model is constructed based on multi-source data of the engineering machinery industry, the strategy income relationship of each industry subject under different market scenarios is analyzed through the competition pattern change model, and the change path of the competition pattern is predicted;

[0042] A target virtual market environment is constructed based on the predicted competition pattern change path through a scene simulation environment model;

[0043] In the target virtual market environment, the income change of different optimization actions in the multi-market scenario is simulated through a reinforcement learning optimization algorithm, the strategy decision parameters are iteratively adjusted according to the simulation results multiple times, and an optimization scheme is determined.

[0044] Further, after determining the optimization scheme, the optimization scheme is executed, and parameters of the competition pattern change model, the reinforcement learning optimization algorithm and the scenario simulation environment model are adjusted based on the execution effect of the optimization scheme.

[0045] Further, the process of constructing the competition pattern change model based on the multi-source data of the engineering machinery industry specifically includes:

[0046] Based on the multi-source data of the engineering machinery industry, historical strategy behavior data, market transaction records, supply chain response data and external macroeconomic indicators of each subject in the industry chain are analyzed as a first data analysis result.

[0047] According to the first data analysis result, a state transition matrix and a strategy benefit matrix are constructed; the state transition matrix is used to describe the change law between different market scenarios, and the strategy benefit matrix represents the strategy selection of each subject in different market scenarios and the change relationship of the benefits.

[0048] Based on the state transition matrix and the strategy benefit matrix, the behavior probability distribution of the subject is updated, and the change path of the industry competition pattern in the future time period is determined according to the behavior probability distribution of the subject.

[0049] Specifically, in the embodiment of the present application, before constructing the competition pattern change model, multi-source data of the engineering machinery industry is collected. The multi-source data of the engineering machinery industry includes but is not limited to historical strategy behavior data, market transaction records, supply chain response data and external macroeconomic indicators of each subject in the industry chain. The historical strategy behavior data reflects the strategy selection trajectory of the enterprise in different market environments, such as capacity expansion, market exit, price adjustment, etc.; the market transaction records include the sales volume, transaction price and transaction frequency of the enterprise in different regions, describe the market activity and the competition strength of the subject; the supply chain response data reflects the supply and demand response time, inventory fluctuation, etc. of the upstream and downstream of the industry chain, reflects the dynamic coordination ability of the industry chain; the external macroeconomic indicators include GDP growth rate, industry investment scale, policy adjustment, etc., which are used to comprehensively consider the influence of external environment on the industry pattern. Compared with other industry data, the engineering machinery industry data has stronger time sequence, periodicity and policy sensitivity, which is the basis for setting the competition pattern change model for this industry.

[0050] Based on the above data, multi-dimensional analysis is performed to extract key variable relationships and form a first data analysis result. This result reflects the potential coupling characteristics of the strategy behavior of each subject and market dynamics. According to the first data analysis result, a state transition matrix is constructed to describe the change law between different market scenarios, where a market scenario refers to a market state formed under a specific demand level, supply chain state, and competitor strategy combination; and the elements of the state transition matrix represent the probability distribution of transitioning from one market state to another. At the same time, a strategy benefit matrix is constructed, which represents the relationship between the benefits of each subject choosing different strategies under different market scenarios, including market share improvement, cost optimization, and other comprehensive indicators.

[0051] Based on the above state transition matrix and strategy benefit matrix, the behavior probability distribution of each subject is updated to reflect the selection tendency of the subject under different market scenarios and different strategies, and according to the updated behavior probability distribution of the subject, the change path of the construction machinery industry competition pattern in the future time period is predicted, i.e., the possible market state evolution sequence and the subject strategy adjustment sequence are generated.

[0052] Further specifically, the updated behavior probability distribution of the subject is taken as input, and preferably, the change path of the industry competition pattern in the future time period is predicted by Markov Chain Monte Carlo (MCMC) simulation, multi-time step prediction algorithm, or dynamic deduction method based on time series. The change path refers to the evolution trajectory of the overall market state of the industry chain and the strategy selection of each subject under multiple market scenarios over time. This path not only includes the market state sequence (such as demand upsurge, supply shortage, and intensified competition), but also contains the specific strategy adjustment sequence made by each subject at different time nodes, reflecting the evolution process of the industry ecosystem.

[0053] In this embodiment, the change path of the construction machinery industry competition pattern aims to describe the dynamic evolution of the entire industry chain structure and enterprise behavior over a period of time, influenced by multiple factors such as market demand changes, supply chain fluctuations, macroeconomic policy adjustments, and competitor responses. This path not only helps industry managers to anticipate market trends, but also provides valuable prior information for subsequent target virtual market environment construction and reinforcement learning optimization algorithms, ensuring that the generated optimization scheme can effectively adapt to complex and changing market environments.

[0054] Through the setting of the competition pattern change model, the present application can achieve high-precision prediction of the evolution process of the construction machinery industry competition pattern in a complex market dynamic environment, providing data basis and theoretical support for subsequent optimization strategy generation and decision support, significantly improving the scientificity and adaptability of industry competitiveness evaluation and prediction.

[0055] Further, the virtual market environment is used to simulate industry chain dynamics, market demand changes, and competitor strategy behavior responses.

[0056] Further, constructing the target virtual market environment based on the predicted competition pattern change path in the scenario simulation environment model includes:

[0057] Extracting industry chain dynamic parameters, market demand fluctuation parameters, and competitor strategy behavior response parameters based on the predicted competition pattern change path;

[0058] Fusing the parameters in the scenario simulation environment model to generate a virtual market environment;

[0059] Adjusting the boundary parameters of the virtual market environment through environmental constraint conditions to obtain a target virtual market environment.

[0060] Specifically, in the embodiments of the present application, the industry chain dynamic parameters describe the resource flow, supply and demand adjustment of the upstream and downstream of the industry chain in the market change process, such as raw material supply speed, logistics distribution cycle, and upstream and downstream enterprise capacity adjustment amplitude; the market demand fluctuation parameters reflect the market demand scale and its change trend in different periods and different regions, including but not limited to seasonal engineering demand, peak period equipment procurement behavior, etc.; the competitor strategy behavior response parameters refer to the possible strategy adjustment of the main competitor when the market state changes, such as price war, product innovation, market expansion, etc.

[0061] Fusing the above parameters in the scenario simulation environment model to generate a preliminary virtual market environment. The virtual market environment refers to a market ecosystem reconstructed in a computer simulation environment, which is used to dynamically simulate the interaction process between industry chain subjects, market demand, and competitor strategies. In the virtual market environment, by setting multiple market scenarios and state switching logic, complex market dynamics can be reproduced, including supply chain bottlenecks caused by sudden demand surge, market share fluctuations caused by competitor strategy conflicts, etc.

[0062] At the same time, the boundary parameters of the virtual market environment are adjusted based on the environmental constraint conditions to generate a target virtual market environment. The adjusted virtual market environment not only accurately reflects complex market dynamics, but also has strong scalability and adaptability.

[0063] In the present application, the target virtual market environment aims to provide a training platform for reinforcement learning optimization algorithms, so that the optimization strategy can be simulated, verified, and dynamically adjusted in multiple market scenarios, ensuring the effectiveness and adaptability of the final output optimization scheme in the real industry environment.

[0064] Further, adjusting the boundary parameters of the virtual market environment through environmental constraint conditions includes:

[0065] The boundary parameters are dynamically adjusted based on market supply and demand balance constraints, industry chain resource utilization constraints and competitive strategy game constraints through a multi-objective optimization algorithm to minimize the deviation between the virtual market environment and the actual market state.

[0066] Specifically, after generating a preliminary virtual market environment based on the predicted competitive pattern change path, multi-dimensional environmental constraint conditions are set to guide the dynamic adjustment of the boundary parameters. The environmental constraint conditions include but are not limited to market supply and demand balance constraints, industry chain resource utilization constraints and competitive strategy game constraints. The market supply and demand balance constraints are used to ensure that the virtual market environment can maintain overall supply and demand coordination during simulation, avoiding extreme situations of resource surplus or shortage; the industry chain resource utilization constraints are used to limit the production capacity, logistics response and upstream and downstream synergy efficiency of each main body in the industry chain, ensuring that the simulation results conform to the actual resource distribution and utilization; the competitive strategy game constraints are used to reflect the dynamic balance relationship between the strategy interaction of the main competitors, avoiding unrealistic strategy dominance or antagonism in the virtual market environment.

[0067] The boundary parameters of the virtual market environment are iteratively adjusted through a multi-objective optimization algorithm. Specifically, a comprehensive deviation minimization objective function is constructed, and the market supply and demand balance error, industry chain resource utilization deviation and competitive strategy game equilibrium degree are used as constraint conditions of the multi-objective function. Further optionally, the boundary parameters are dynamically adjusted through optimization strategies such as weight allocation or Pareto optimal solution search. During the optimization process, the boundary parameters such as resource capacity upper limit, market entry threshold and strategy adjustment flexibility are continuously corrected in iteration until the difference between the overall state of the virtual market environment and the actual market state is reduced to within the preset threshold, achieving high fitting degree between the virtual environment and the real market.

[0068] In the embodiments of the present application, the boundary parameters of the virtual market environment are adjusted through a multi-objective optimization algorithm, which not only improves the simulation accuracy of the virtual market environment in complex scenarios, but also effectively supports the training and strategy verification of the subsequent reinforcement learning optimization algorithm, so that the generated optimization scheme can adapt to the real market dynamics, achieving high consistency and practicality of industry competitiveness evaluation and optimization.

[0069] Further, before constructing the scenario simulation environment model, the method further comprises:

[0070] An environment factor set is constructed based on multi-source data of the engineering machinery industry; the environment factor set includes market demand, industry chain resource constraints and competitive opponent strategy behavior;

[0071] The weight proportion of each environment factor is adjusted according to historical data and competitive pattern change path prediction results.

[0072] Further, the construction process of the scenario simulation environment model comprises:

[0073] According to the environment factor set and the weight proportion, a transition relationship between different market scenarios is determined, and a scenario simulation environment model is constructed based on the environment factor set, the weight proportion and the transition relationship between different market scenarios.

[0074] Specifically, in the embodiment of the present application, the environment factor set refers to a combination of key parameters affecting the dynamic change of the engineering machinery industry market, including but not limited to market demand, industry chain resource constraints and competitor strategy behavior, etc. core factors. Market demand reflects the change trend of engineering equipment procurement and construction project starting scale in different regions and different time periods, and is an important factor driving the change of market state; the industry chain resource constraint covers the raw material supply capacity, production and logistics efficiency, etc., and reflects the resource flow restriction of the upstream and downstream of the industry chain; the competitor strategy behavior represents the market expansion, price adjustment and product innovation, etc. response strategies that the main enterprises may take in different market situations, and describes the influence of the game behavior in the industry on the market pattern. The purpose of optimizing the three types of environment factors as the core of the set is that they together constitute the dominant variable of the evolution of the engineering machinery industry market state, and can comprehensively reflect the dynamic characteristics of the industrial ecosystem.

[0075] The influence intensity of each environment factor in the past market change can be obtained through historical data, and the competition pattern change path prediction result reveals the dynamic change trend of the action of each factor in the future potential market state. On this basis, the weight proportion is adjusted, so that the model can adapt to complex and variable market conditions, and avoid distortion of prediction caused by excessive dominance of a factor.

[0076] Further specifically, when adjusting the weight proportion of each environment factor, preferably, multivariate linear regression analysis, principal component analysis (PCA) or information gain calculation, etc. are used to extract the contribution of market demand, industry chain resource constraint and competitor strategy behavior to the evolution of market state, and generate a preliminary weight distribution. At the same time, the competition pattern change path prediction result is introduced into the analysis process, and the preliminary weight is dynamically adjusted by Bayesian updating algorithm, so as to ensure that the weight proportion not only reflects the historical influence trend, but also can integrate the dynamic action change of each factor in the future potential market state, so that the finally obtained weight proportion balances the historical experience data and has forward-looking prediction ability, avoiding distortion of the model caused by excessive dominance of a factor.

[0077] Next, based on the adjusted weight set of environmental factors, the transition relationship between different market scenarios is determined. Specifically, a state transition probability matrix is used, where the matrix elements are set to represent the probability value of transitioning from the current market state (such as high demand, resource shortage, and intensified competition) to the next market state (such as medium demand, resource balance, and relaxed competition). The probability value is calculated based on the Markov chain assumption, combined with the weight proportion of environmental factors to adjust the state transition path, ensuring the logical continuity and complexity of scenario changes.

[0078] Finally, the scenario simulation environment model is constructed by integrating the set of environmental factors, weight proportions, and market scenario transition relationships. Through this model, the interactive process of industry chain subjects, market demand fluctuations, and competitor strategy responses in a computer simulation environment is dynamically simulated. Further preferably, a dynamic boundary parameter adjustment unit is embedded in the model for real-time correction of resource capacity, market entry threshold, and strategy flexibility, enabling approximation of the multi-market scenario set of the real market environment.

[0079] In the embodiment of the application, the scenario simulation environment model realizes the visual reproduction and dynamic deduction of the complex market dynamics of the engineering machinery industry, providing a training and verification platform for subsequent reinforcement learning optimization algorithms, ensuring that the optimization strategy can adapt to changing markets and achieve dynamic improvement of industrial competitiveness.

[0080] Further, the reinforcement learning optimization algorithm simulates the changes in the benefits of different optimization actions in the multi-market scenario in the target virtual market environment, which specifically includes:

[0081] Building a state space, action space, and reward mechanism based on the target virtual market environment;

[0082] Under the constraints of the constructed space and reward mechanism, the trained policy network simulates the changes in the benefits of each optimization action in the multi-market scenario and builds an action benefit mapping relationship;

[0083] Adjusting the strategy selection logic in the reinforcement learning optimization algorithm based on the action benefit mapping relationship.

[0084] Specifically, the state space refers to the environmental state set of the engineering machinery industry market at different time nodes, representing key variables that affect strategy decision-making, including but not limited to market demand intensity, industry chain resource utilization rate, competitor strategy dynamics, policy regulation factors, and other key variables. The state space is represented by a multi-dimensional vector, defined as S t = [[d t ,r t ,b t ,p t ], where d t represents the current market demand index, r tdenotes resource utilization, b t denotes competitor strategy game parameters, p t denotes policy adjustment factor.

[0085] The action space refers to the set of optimization strategies that can be selected under each market state, including capacity allocation (such as expanding production by 10%, reducing production by 20%), product portfolio adjustment (such as increasing the proportion of high-value-added products, eliminating low-profit products), market entry path selection (such as entering emerging markets, exiting saturated markets), etc. The action space is defined as A t ={a1, a2,..., a j ,...,a k}, where each a j represents an executable optimization action.

[0086] The reward mechanism is used to evaluate the long-term value of each action under a given state, considering not only profit maximization but also resource constraints and risk factors, and is defined as follows:

[0087] R(S t ,A t ) = α·Δπ t - β·C t - γ·R f

[0088] where Δπ t represents the profit gain brought by the optimization action A t ; C t represents the additional cost caused by resource scheduling; R f represents the market risk penalty factor; α, β, γ are weight coefficients, and Σα, β, γ = 1; t is the time step.

[0089] Through this setting, it is ensured that the reinforcement learning process not only focuses on profit maximization, but also considers resource consumption and market risk, which conforms to the actual operation logic of the engineering machinery industry and provides a multi-objective decision basis, so that the optimization strategy can achieve profit growth while reducing supply chain burden and market risk.

[0090] On this basis, the strategy function is constructed to output the probability distribution of selecting action A t under state S t :

[0091] π θ (S t ,C) = P(A t |S t ,C)

[0092] Wherein C represents environmental constraints, including market supply and demand balance constraints, industry chain resource utilization constraints and competition strategy game constraints, for ensuring the feasibility of strategy selection; θ is a strategy network parameter, which is updated through multiple rounds of iterative optimization.

[0093] As a preferred embodiment, the strategy network is implemented by a deep neural network, and a policy gradient algorithm is used for training to maximize the cumulative reward:

[0094]

[0095] Wherein γ is a discount factor, used to balance long-term and short-term returns.

[0096] The above formula is the core optimization objective of reinforcement learning, which drives the continuous updating of the strategy network parameter θ, so that the long-term optimal action selection strategy can be learned in the virtual market environment, avoiding short-sighted behavior and improving the overall competitiveness of the industry.

[0097] As a further preferred embodiment, in the training process, an entropy regularization term H(π θ ) is introduced, and an exploratory adjustment function λ is introduced to enhance the exploratory nature of the strategy and prevent premature convergence to a local optimal solution, and the overall optimization objective function is L(θ):

[0098] L(θ)=J(θ)+λH(π θ )

[0099] Through the entropy regularization mechanism, the strategy network can maintain sufficient randomness in the early training stage to avoid premature convergence to a local optimal solution, so that potential high-yield strategy paths can be found in the complex and variable construction machinery market, enhancing the applicability and innovation of the optimization scheme.

[0100] Further, the iterative adjustment of the strategy decision parameters and the determination of the optimization scheme based on the simulation results include:

[0101] Based on the adjusted strategy selection logic, the strategy parameters are optimized through multiple rounds of training operations;

[0102] According to the training results, a preliminary optimized action sequence is generated in the target virtual market environment;

[0103] After executing the preliminary optimized action sequence, the sequence is adjusted based on simulation feedback, maximum return constraints, resource constraints and market risk factors to generate a final optimized action sequence;

[0104] Wherein the final optimized action sequence is used to determine the optimization scheme including capacity allocation, product portfolio adjustment and market entry path.

[0105] It can be understood that the strategy selection logic has a preliminary decision-making ability after the preliminary reinforcement learning training, but due to the complex and changeable market environment of the engineering machinery industry, relying only on one training is easy to cause the strategy network to perform poorly in some edge scenarios, or unable to fully capture the dynamic interaction of the upstream and downstream of the industry chain. Therefore, in the embodiment of the present application, through multiple rounds of training operations, combined with the simulation results of different market scenarios, the strategy network parameters are iteratively updated, so that it has stronger adaptability to diversified market environment, and avoids the strategy falling into a local optimal solution.

[0106] Subsequently, according to the optimized strategy network, a preliminary optimization action sequence is generated in the target virtual market environment. The optimization action sequence refers to a set of executable optimization actions arranged in time sequence, each optimization action corresponding to a specific adjustment strategy under a specific market state, for example, expanding production capacity in a market demand boom state, optimizing resource allocation in a supply chain bottleneck state, and adjusting product portfolio in a intensified competition state. The sequence not only considers short-term revenue, but also combines long-term market dynamics to evaluate its cumulative revenue within the entire prediction period through multi-scenario simulation results.

[0107] After executing the preliminary optimization action sequence, fine-tuning is performed based on simulation feedback. Specifically, the sequence is dynamically adjusted through revenue maximization constraints, resource utilization constraints, and market risk factors to optimize the action plan to avoid resource overload, reduce potential risks, and further improve revenue. This process preferably uses a multi-objective optimization algorithm to search for a Pareto optimal solution set under the premise of meeting supply and demand balance and market stability, and determine the optimization action sequence that balances revenue and risk. The finally generated optimization action sequence includes decision-making information such as production capacity allocation, product portfolio adjustment, and market entry path for the engineering machinery industry, which is used to support enterprises to make scientific competitive enhancement decisions in complex market environment.

[0108] Through the above multiple rounds of strategy training, optimization action sequence generation, and dynamic adjustment based on simulation feedback, an optimization scheme that can be deployed in the real industry is learned from the virtual environment, ensuring that the optimization strategy not only has theoretical feasibility, but also can adapt to complex and changeable market dynamics, and improve the overall competitiveness of the engineering machinery industry.

[0109] Finally, it should be noted that the above-described embodiments include a plurality of parallel embodiments of the present application, and deleting or otherwise adjusting one or more embodiments does not affect the implementation of the scheme. In addition, the disclosed engineering machinery industry competitiveness evaluation and optimization method based on multi-dimensional analysis disclosed by the embodiments of the present application is only the preferred embodiment of the present application, and is only used to illustrate the technical solutions of the present application, but not to limit it; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand; the technical solutions recorded in the foregoing embodiments can be modified, or some technical features can be replaced; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for evaluating and optimizing the competitiveness of the construction machinery industry based on multi-dimensional analysis, characterized in that, The method includes: Based on multi-source data from the construction machinery industry, a competitive landscape change model is constructed. The model is used to analyze the strategic benefit relationships of various industry players in different market scenarios and predict the path of competitive landscape change. A target virtual market environment is constructed based on the predicted path of competitive landscape change through a scenario simulation environment model; In the target virtual market environment, the algorithm of reinforcement learning is used to simulate the changes in returns of different optimization actions in multiple market scenarios. Based on the simulation results, the strategy decision parameters are adjusted multiple times to determine the optimization scheme.

2. The method for evaluating and optimizing the competitiveness of the construction machinery industry based on multi-dimensional analysis according to claim 1, characterized in that, After determining the optimization scheme, the optimization scheme is executed, and the parameters of the competitive landscape change model, reinforcement learning optimization algorithm, and scenario simulation environment model are adjusted based on the execution effect of the optimization scheme.

3. The method for evaluating and optimizing the competitiveness of the construction machinery industry based on multi-dimensional analysis according to any one of claims 1-2, characterized in that, The process of constructing a competitive landscape change model based on multi-source data from the construction machinery industry specifically includes: Based on multi-source data analysis of the construction machinery industry, the historical strategic behavior data of various entities in the industrial chain, market transaction records, supply chain response data, and external macroeconomic indicators are used as the primary data analysis results. Based on the results of the first data analysis, a state transition matrix and a strategy payoff matrix are constructed. The state transition matrix is ​​used to describe the changing patterns between different market scenarios, and the strategy payoff matrix represents the strategy choices and payoff changes of each entity in different market scenarios. The probability distribution of the subject's behavior is updated based on the state transition matrix and the strategy payoff matrix, and the path of change in the industry's competitive landscape in the future time period is determined based on the probability distribution of the subject's behavior.

4. The method for evaluating and optimizing the competitiveness of the construction machinery industry based on multi-dimensional analysis according to any one of claims 1-2, characterized in that, The virtual market environment is used to simulate industry chain dynamics, changes in market demand, and competitor strategies and responses.

5. The method for evaluating and optimizing the competitiveness of the construction machinery industry based on multi-dimensional analysis according to claim 4, characterized in that, The construction of the target virtual market environment based on the predicted competitive landscape change path through scenario simulation environment model specifically includes: Based on the predicted competitive landscape change path, dynamic parameters of the industrial chain, market demand fluctuation parameters, and competitor strategy and behavior response parameters are extracted. The parameters are integrated into the scenario simulation environment model to generate a virtual market environment; The target virtual market environment is obtained by adjusting the boundary parameters of the virtual market environment through environmental constraints.

6. The method for evaluating and optimizing the competitiveness of the construction machinery industry based on multi-dimensional analysis according to claim 5, characterized in that, The adjustment of boundary parameters of the virtual market environment through environmental constraints includes: By employing a multi-objective optimization algorithm based on market supply and demand balance constraints, industrial chain resource utilization constraints, and competitive strategy game constraints, the boundary parameters are dynamically adjusted to minimize the deviation between the virtual market environment and the actual market state.

7. The method for evaluating and optimizing the competitiveness of the construction machinery industry based on multi-dimensional analysis according to claim 6, characterized in that, Before constructing the scene simulation environment model, the method further includes: An environmental factor set is constructed based on multi-source data from the construction machinery industry; the environmental factor set includes market demand, supply chain resource constraints, and competitor strategies and behaviors. The weighting of each environmental factor is adjusted based on historical data and predictions of changes in the competitive landscape.

8. The method for evaluating and optimizing the competitiveness of the construction machinery industry based on multi-dimensional analysis according to claim 7, characterized in that, The process of constructing the scenario simulation environment model includes: The transfer relationships between different market scenarios are determined based on the set of environmental factors and their weight ratios, and a scenario simulation environment model is constructed based on the set of environmental factors, their weight ratios, and the transfer relationships between different market scenarios.

9. The method for evaluating and optimizing the competitiveness of the construction machinery industry based on multi-dimensional analysis according to any one of claims 1-2, characterized in that, The specific steps of simulating the changes in returns under multiple market scenarios through reinforcement learning optimization algorithms in a target virtual market environment include: Construct a state space, action space, and reward mechanism based on the target virtual market environment; Under the constraints of the constructed space and reward mechanism, the changes in the revenue of each optimization action are simulated in multiple market scenarios based on the trained policy network, and the action revenue mapping relationship is constructed. The strategy selection logic in the reinforcement learning optimization algorithm is adjusted based on the aforementioned action-reward mapping relationship.

10. The method for evaluating and optimizing the competitiveness of the construction machinery industry based on multi-dimensional analysis according to claim 9, characterized in that, The step of iteratively adjusting the strategy decision parameters based on simulation results and determining the optimization scheme specifically includes: Based on the adjusted strategy selection logic, the strategy parameters are optimized through multiple rounds of training. Based on the training results, generate a preliminary optimized action sequence within the target virtual market environment; After executing the initial optimized action sequence, the sequence is adjusted based on simulation feedback, profit maximization constraints, resource constraints, and market risk factors to generate the final optimized action sequence. The final optimized action sequence is used to determine the optimized scheme, including capacity allocation, product portfolio adjustment, and market entry path.

Images