Innovation of Enterprise Management Mode Based on Improved

RBF Neural Network Algorithm

Zhang Jing

Guangdong Vocational Institute Of Public Administration, Zhongluotan Town, Baiyun District, Guangzhou City,

Guangdong Province, 510545, China

Keywords: Improved RBF Neural Network Algorithm, Enterprise Management Model, Improvement, Innovation.

Abstract: In a rapidly changing market environment, companies are facing increasing competitive pressure. In order to

improve operational efficiency, optimize resource allocation and improve the quality of decision-making, this

study explores the application of improved RBF neural network algorithm in enterprise management model

innovation. Based on the comprehensive analysis of the internal management data of the enterprise, the

management decision-making model and system based on RBF neural network are studied and constructed,

and optimized and tested. The final experimental data show that the model integration system has significantly

improved the prediction accuracy and response speed, and the comprehensive conclusion shows that the

improved RBF neural network algorithm can effectively support the innovation of enterprise management

mode and realize intelligent and refined management.

1 INTRODUCTION

In the current competitive business environment, how

to improve the management efficiency and decision-

making quality of enterprises has become an

important topic. Traditional management methods

rely on experience and intuition, and are unable to

cope with complex and changing market conditions.

Some people have proposed methods based on data

analysis, but because the data is too dimensional and

too complex, these methods cannot achieve ideal

results in practical application. There are also

attempts to adopt a rules-based management system

for this application. However, these methods cannot

effectively deal with nonlinearity and uncertainty,

resulting in poor management effectiveness. In this

paper, the improved RBF neural network algorithm is

used to study the innovation of enterprise

management mode, because the algorithm has good

nonlinear processing ability, can adapt to the complex

enterprise management environment, and at the same

time, based on intelligent optimization and machine

learning, it can also improve the accuracy and

efficiency of management decision-making. It is

hoped that based on this study, a new enterprise

management model can be explored, intelligent and

data-driven decision support can be realized, and the

core competitiveness of enterprises can be enhanced.

2 RELATED WORKS

2.1 RBF Neural Network Algorithm

Theory

RBF neural network is a commonly used artificial

neural network model, which is mainly used to deal

with nonlinear problems and function approximation

problems. The core of the RBF network is based on

the activation method of radial basis function, which

maps the input data to a high-dimensional space, and

uses it to achieve the purpose of nonlinear

approximation of complex relations (Cao, and Yu.

2023). The RBF neural network consists of three

layers, specifically, they are the input layer, the

hidden layer, and the output layer. The function of the

input layer is to receive external data, and the input

variables represent the various characteristics of the

problem; These characteristics often represent

various management elements in enterprise

management innovation, such as financial data,

market dynamics, etc. The hidden layer uses the

radial basis function to perform nonlinear

transformation on the input data, and the optimization

of the innovation investment of the hidden layer is

Jing, Z.

Innovation of Enterprise Management Mode Based on Improved RBF Neural Network Algorithm.

DOI: 10.5220/0013535500004664

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 3rd International Conference on Futuristic Technology (INCOFT 2025) - Volume 1, pages 59-65

ISBN: 978-989-758-763-4

very important in enterprise management, which

determines the complexity and computational

accuracy of the network. The task of the output layer

is to map the output of the hidden layer to the final

result, which is usually used to represent the decision-

making suggestions and performance indicators in the

innovation of enterprise management models

(Chaynikov, Semenov, et al. 2023). The network

learns the relationship between input and output

based on adjusting weights and center vectors, which

is suitable for modeling and optimizing multi-factor

decision-making problems in enterprise management.

2.2 Theories Related to the Innovation

of Enterprise Management Model

Enterprise management model innovation refers to

the process of optimizing management processes,

adjusting organizational structure, and introducing

advanced management tools in the face of external

market changes and insufficient internal operational

efficiency, so as to effectively improve overall

efficiency and optimize strategy (Chaynikov,

Semenov, et al. 2023). The core theories of

innovation management model include resource base

theory, dynamic capability theory and change

management theory. Among them, the resource-

based theory believes that the competitive advantage

of an enterprise comes from the effective allocation

of its internal resources and the mining of its unique

capabilities. In the innovation of enterprise

management mode, it can enhance the overall

competitiveness of enterprises based on the optimal

allocation of production resources, human resources,

and technical resources (Gao, and Yang. 2023).

Dynamic capability theory emphasizes the ability of

enterprises to adapt and innovate in a dynamic

environment. Enterprise management model

innovation should have the ability to quickly respond

to market changes and continue to innovate, based on

intelligent management tools, such as RBF neural

network algorithms, to effectively improve its speed

and adaptability. Change management theory, which

mainly explores how to carry out organizational

change and strategic adjustment in the face of

changes in the internal and external environment. In

the process of innovation of its management model, it

can promote the further sustainable development of

the enterprise by adjusting the corporate structure,

optimizing the management process, and stimulating

the creativity of employees. Based on the theory of

combining RBF neural network algorithm with

enterprise management model innovation (Kang,

Zhao, et al. 2023), enterprises can achieve accurate

data analysis and intelligent decision support in the

process of management change. to promote

enterprises to maintain competitive advantage and

sustainable development in the fierce market

competition.

3 METHODS

3.1 Introduction to the Innovation

System of Enterprise Management

Mode

In the system architecture, each part performs its own

duties and cooperates with each other to achieve

intelligent and efficient enterprise management

decision-making. The key modules in the system

architecture are the data preprocessing module, the

input layer module, the hidden layer module, the

output layer module, the model training module, the

model optimization module, and the evaluation and

feedback module (Tang, and Yang. 2023). Among

them, the task of the data preprocessing module is to

process all aspects of the original management data,

including cleaning, so as to ensure the quality and

consistency of the data related to the innovation of

enterprise management mode. The module also

performs feature extraction and feature selection to

improve the efficiency and effectiveness of model

training. The input layer module is tasked with

receiving key variables in enterprise management,

such as financial data, market indicators and

production efficiency. The input layer converts these

variables into numeric forms for processing by the

neural network. The task of the hidden layer module

is nonlinear feature extraction and pattern

recognition. Based on adjusting the input and weight

of innovation, the module is able to deal with

complex management decision-making problems.

The task of the output layer module is to convert the

output of the hidden layer into specific management

decision suggestions, such as resource allocation

plans and market strategy adjustments. The module

also outputs model prediction results for

management's reference. The task of the model

training module is to train the neural network using

historical data and simulation data. Based on

adjusting parameters, such as weights and learning

rate, the prediction accuracy and generalization

ability of the model are optimized. The task of the

Model Optimization module is to improve the kernel

function, learning rate, and regularization method of

the model to further improve the performance of the

INCOFT 2025 - International Conference on Futuristic Technology

model in different management scenarios. The task of

the Evaluation & Feedback module is to evaluate the

output of the model and provide feedback based on

the actual results. This module helps to continuously

adjust and optimize the model to improve the

scientific and accurate decision-making. These

modules work together to innovate and optimize the

enterprise management model, enabling enterprises

to respond to market changes and management

challenges faster and more effectively (Wang, Tian,

et al. 2023).

3.2 Enterprise Management Model

Innovation System Design

Model building is the key to enterprise model

innovation. Specifically, the process is:

Design the input layer. The main task of the input

layer is to obtain key variables from enterprise

management, which mainly include financial

indicators, market sales data, production efficiency,

human resource allocation, etc. See Eq. (1) for details.

]X[x,x,,x

=…

)

In this formula, the input vector represents the

multi-dimensional factors in the enterprise

management model, for example,

is refers to the

market share of the enterprise, and represents the

competitiveness and brand influence of the enterprise

in the market;

is refers to the production cost,

which reflects the cost control ability of the

enterprise;

is employee satisfaction, which affects

the human resource efficiency and innovation ability

of the enterprise. These input data will be

standardized, and based on this, it will be used to

ensure that variables of different dimensions are used

in the model reasonably, and then ensure the data

uniformity and the accuracy of the model.

It is also necessary to optimize the investment in

innovation in the hidden layer. Specifically, the

hidden layer is the innovation investment of RBF

neural network, which directly affects the

computational performance and learning ability of the

model. In the innovation of enterprise management

mode, the nodes of the hidden layer need to be

adjusted according to the needs of enterprise

management. For the formula, see Eq. (2).

hn=α×

)

In this formula,

is the innovation input of the

hidden layer is represented.

is an evaluation index

that determines the flexibility of the amount of

innovation input in the hidden layer. Generally

speaking, cross-validation is required to make this

determination.

is the number of input variables that

directly reflects the breadth of key factors considered

in business management. Reasonable innovation

investment can help the model achieve rapid response

and efficient calculation when dealing with complex

problems in enterprise management (Wang, 2024),

and then avoid overfitting or computational

redundancy due to excessive innovation investment.

Determining its output layer mapping function is

another key step. The main task of the output layer is

to translate the calculation results of the hidden layer

into specific enterprise management decisions, and its

mapping function form is shown in Eq. (3).

yw(||XC||)

=×φ−



)

In this formula,

is the weight, which

represents the degree of influence of each factor on

the final decision in the innovation of enterprise

management mode. For example, the greater weight

of market share indicates that it is more important in

corporate decision-making. Weights are continuously

adjusted based on training to optimize decision-

making.

is an activation function that is used to

measure the degree of matching between input

variables and central vectors in the management of

enterprises. The improved Gaussian activation

function can be used to deal with the nonlinear

changes in market demand and supply chain

fluctuations in management.

3.3 Research and Training on

Enterprise Management Model

Innovation

In this paper, the model training performs supervised

learning by introducing historical data and simulation

data of enterprises, and continuously optimizes the

weights and parameters of the RBF neural network.

Based on multiple iterations and parameter

adjustments, the model gradually improves its

prediction ability for complex management scenarios.

In the training process, it also realizes the rapid

convergence of the model based on the error feedback

mechanism to ensure the decision support of the

Innovation of Enterprise Management Mode Based on Improved RBF Neural Network Algorithm

management side under different enterprise

management modes (Xiao, 2024).

The model optimization is based on adjusting the

kernel function, learning rate and regularization

method to further improve its adaptability to the

innovation of enterprise management mode. Among

them, the kernel function is the key for the model to

identify nonlinear patterns in enterprise management.

For the improved Gaussian kernel function, see Eq.

(4).

|| X - C ||

f(|| X - C ||)= exp -

2σ







)

n Eq. (4),

is the Gaussian kernel width

parameter, which reflects the sensitivity of the model

to the distribution of input data. In the enterprise

management model, the adjustment will effectively

optimize the performance of the model in different

business scenarios, such as in a volatile market

environment, which can improve the robustness.

The adaptive learning rate mechanism can make

the improved RBF model gradually optimized in

enterprise management decision-making. For the

learning rate, the formula is shown in (5).

t+1 t

η = η ×

1+ β×t

)

In this formula,

is the current learning rate,

which is used to control the rate at which the

parameters of the model are updated. This parameter

is large in the early stage to speed up the model

learning, but gradually decreases in the later stage to

ensure stable convergence.

is a tuning parameter

that has a direct effect on the rate of decay of the

learning rate (Yan, and Ma, 2023). In the innovation

of enterprise management mode, based on the fine

adjustment of the learning rate, the rapid adaptation

of the model in changing management scenarios will

be further ensured.

Regularization is mainly to constrain the

complexity of the model and avoid overfitting under

the operation of the penalty term in the loss function,

so that the results of enterprise management model

innovation are more universal. For the formula, see

Eq. (6).

2h2

i=1

L= (y -yi) +λ i=1 w



)

In this formula,

is the fitting ability, which is

used to control the size of the weights and prevent the

RBF model from relying too much on specific

variables in enterprise management decisions, so as

to improve the robustness of the decision. Based on

the improved RBF neural network algorithm model,

the innovation of enterprise management mode will

be realized. It can make scientific decisions in a

complex and changeable environment, provide

intelligent support for enterprises, optimize resource

allocation and strategic choice, and then achieve

continuous improvement and innovation of

management mode.

3.4 Research and Optimization of

Enterprise Management Model

Innovation

In the innovation of enterprise management mode

based on the improved RBF neural network

algorithm, the system integration method needs to

effectively integrate each module to ensure the

smooth processing of data and the seamless

connection of the decision-making optimization

process. System integration involves these steps. For

example, data interface integration, this step is based

on the design of standardized data interfaces to realize

the data interaction between data preprocessing

modules and input layer modules. This process

ensures that the data is transmitted and processed in a

uniform format. The integration of invocation

between modules is mainly to let each functional

module call based on API and microservice

architecture, such as the hidden layer module to call

the processed data. The integration between the

modules is carried out synchronously or

asynchronously to ensure the efficiency of the

system. Algorithm integration, the core algorithm of

the RBF neural network is integrated with the model

training module to ensure that the optimization

algorithm is automatically called to adjust parameters

such as weights, kernel functions, and regularization

methods during the training process. Output

integration, the output layer module integrates the

prediction results, evaluation and feedback modules,

and then effectively evaluates the model output

results, and optimizes the feedback, so that the

management can view and apply these decision-

making suggestions in real time. Based on system

integration, each module can cooperate efficiently

and effectively achieve the overall optimization of

enterprise management mode.

INCOFT 2025 - International Conference on Futuristic Technology

4 RESULTS AND DISCUSSION

4.1 Case Introduction of Enterprise

Management Model Innovation

System

This case study is based on the application of an

improved RBF neural network algorithm in the

management model innovation of a large

manufacturing enterprise. In the highly competitive

market environment, the company faced problems

such as rising production costs, complex inventory

management, and fluctuating market demand, The

key data indicators in the overall management mode

of the enterprise are shown in Table 1.

Table 1: Key data indicators of the company

Project

Short

term

profitabi

lity of

the

enterpri

se.

Long

term

econom

benefits

of the

professi

on.

Resource

managem

ent.

The

develop

ment

strategy

of the

enterpris

Product

ion

costs

200,000

210,00

195,000

220

thousand

Sales 500,000

480

thousan

520,000 495,000

Invento

levels

1000

pieces

900

pieces

1050

pieces

950

pieces

Based on Table 1, it can be seen that the company

has large fluctuations in cost control and inventory

management, which has an impact on its overall

operational efficiency.improve the operational

efficiency of the enterprise, optimize the allocation of

resources, and improve the market responsiveness,

the enterprise decided to integrate the system based

on the introduction of the algorithm, so as to optimize

its management model. A total of 400 people

participated in this system test, which was held for 5

days, producing 1,000 products and selling 500

products. The memory analysis of the management

mode data of the enterprise is shown in Figure 1.

Figure 1: Cluster analysis of enterprise management data.

4.2 The Overall Innovation Model of

the Enterprise

The allocation of human resources of enterprises is

not really synchronized with the fluctuations of

market demand, resulting in low efficiency of human

input in production departments and sales

departments. Based on the improved RBF neural

network model integration system, it can effectively

analyze these data of the enterprise and realize the

intelligent optimization of the management mode.

The analysis of the management effectiveness of the

enterprise is shown in Table 2.

Table 2: Analysis of changes in market demand

Time

period

The

economi

benefits

of the

enterpris

Human

resource

manageme

nt level.

The

comprehensi

ve potential

development

enterprises.

Enterprise

developme

nt stage.

300

pieces

350 pieces 320 pieces

A self

optimizatio

hase.

450 pcs 470 pieces 440 pieces

Future

developme

nt strategy

of the

enterprise.

500

pieces

510 pcs 495 pieces

Table 2 shows the fluctuations in market demand

in different quarters, and the changes in demand have

an impact on their production planning and inventory

management strategies. The innovation and

development of enterprise management mode are

shown in Figure 2.

Innovation of Enterprise Management Mode Based on Improved RBF Neural Network Algorithm

Figure 2: Comparison of comprehensive development

effects of enterprises.

4.3 The Comprehensive Judgment

Results of the Enterprise

The comprehensive analysis of the above three tables,

it can be seen that there are significant fluctuations in

the company's production costs and inventory

management, and the rise in production costs and the

instability of inventory have led to a decline in its

operational efficiency. At the same time, the market

demand fluctuates greatly, and the company does not

adjust the production plan in time, so it brings about

a backlog of inventory and insufficient supply.

Table 3: Analysis of the distribution and efficiency of

human resources

Department

Manpower

input1

Manpower

input 2

Manpower

input 3

Production

artment

200 people 220 people 210 people

Sales

artment

150 people 145 people 155 people

Logistics

artment

50 people 55 people 48 people

Table 3 provides data on the allocation of human

resources and the efficiency of departments, and

shows how the staffing and efficiency of each

department in the company have changed over time.

The analysis of the degree of innovation in the

management mode of enterprises is shown in Figure

Figure 3: Comprehensive innovation of enterprise

management mode.

5 CONCLUSIONS

After the study, it can be demonstrated that the

application of the improved RBF neural network

algorithm in the innovation of enterprise management

mode has significant effectiveness. The algorithm can

effectively deal with the complexity problems in

enterprise management, and gives excellent

performance in resource allocation optimization,

production efficiency improvement and decision

support. Based on the introduction of intelligent data

analysis and optimization models, the accuracy and

response speed of enterprise management decision-

making have been greatly improved, so as to achieve

the refinement and intelligence of enterprise

management process. In addition, the algorithm also

has good adaptability in the changing market

environment, which provides strong technical force

for the continuous innovation and competitiveness of

the enterprise. The research time of this paper is quite

limited, so there are inevitably errors and omissions,

which can be further updated and optimized in the

future.

REFERENCES

Cao, D. T., And Y. K. Yu. (2023). Top Management Team

Stability And Enterprise Innovation: A Chairman's

Implicit Human Capital Perspective. Managerial And

Decision Economics, 44(4), 2346-2365.

Chaynikov, V., V. Semenov, A. Suchkova, M. Mityugina,

T. Kravchenko, And O. Chainikova. (2023a).

Approaches To Assessing The Effectiveness Of

Management Of A Machine-Building Enterprise.

Revista Gestao & Tecnologia-Journal Of Management

And Technology, 23, 103-111.

———. (2023b). Approaches To Assessing The

Effectiveness Of Management Of A Machine-Building

Enterprise. Revista Gestao & Tecnologia-Journal Of

Management And Technology, 23(1), 316-324.

Gao, J., And H. Yang. (2023). Practical Application Of Big

Data Statistical Analysis Method For Enterprise

Economic Management In Digital Era. Eai Endorsed

Transactions On Scalable Information Systems, 10(6).

Kang, Z., Y. Zhao, And D. Kim. (2023). Investigation Of

Enterprise Economic Management Model Based On

Fuzzy Logic Algorithm. Heliyon, 9(8).

Tang, W., And S. L. Yang. (2023). Enterprise Digital

Management Efficiency Under Cloud Computing And

Big Data. Sustainability, 15(17).

Wang, K., G. X. Tian, And Y. C. Tao. (2023). Quantitative

Model Of Financial Risk Management Of Forestry

Enterprises Based On Nonlinear Differential Equation.

Journal Of Computational Methods In Sciences And

Engineering, 23(2), 809-823.

INCOFT 2025 - International Conference on Futuristic Technology

Wang, S. C. (2024). The Correlative Factors Of Enterprise

Culture And Enterprise Human Resource Based On

Data Analysis And Federated Learning. Advances In

Continuous And Discrete Models, 2024(1).

Xiao, Z. L. (2024). Investigation On The Strategy Of

Business Administration Level Of Enterprises In The

New Economic Environment. Plos One, 19(5).

Yan, G. Y., And R. Ma. (2023). The Mechanism Of The

Role Of Big Data Knowledge Management In The

Development Of Enterprise Innovation. International

Journal Of Advanced Computer Science And

Applications, 14(6), 1364-1372.

Innovation of Enterprise Management Mode Based on Improved RBF Neural Network Algorithm