Computational model of enterprise cooperative technology innovation risk based on nerve network

Abstract

Collaborative innovation has become the principal innovation method because of the puniness of the innovation strength of the enterprises. Due to the objective reality nature of the risk of enterprise collaborative technology innovation, it is necessary to take measures to prevent and indemnify the loss which the risk may bring. Because there is the complex nonlinear function mechanism between risk factors, the cooperative mode and control mechanism of enterprise collaborative innovation can be studied by nonlinear method. First, this paper analyzed the seeking method of enterprise collaborative innovation risk, and then the concept of controlling risk regulation gradient of the cooperating technological innovation under network environment was explained. And a complete controlling risk model of the cooperating technological innovation has been put forward, which is based on the wavelet and nerve network. Finally, the discussion about the conclusion of the research was given.

Keywords

Risk analysis model cooperative innovation model of controlling risk nerve network wavelet analysis

Introduction

There are existing uncertainties in the process of enterprise collaborative technology innovation, which we cannot completely eliminate and evade, but can take measure to prevent and control the risk.¹ The collaborative innovation of enterprise is a continuous process from the research and development of production to market, and therefore, the method of preventing and managing is necessary to unify the whole innovation process.

Enterprise innovation costs are increasing, and innovation environment becomes more and more difficult; enterprises are usually tough to complete the technological innovation alone. In order to gain market competitive advantage and improve market share, enterprises must establish different forms of strategic cooperation alliance, share all kinds of resources and bring into full play their respective advantages, the implementation of cooperation and innovation. Cooperative innovation refers to enterprises by establishing technological cooperation relationship with other enterprises, scientific research institutions, universities, etc., while maintain the relatively independent of their respective interests and social identities at the same time, and carry out cooperation in research and development of products, technology, or jointly determine the research and development goal of technological innovation activities in a period time.² Cooperative innovation can realize the sharing of resources, take advantage of the complementary relationship innovation cycle, reduce the cost of innovation, so as to enhance the innovation ability of enterprises; thus, it has become the main mode of enterprise innovation. Cooperative innovation usually requires more than two participants.³ This results in the relationship between parties in the enterprise cooperation innovation risk confronting many inevitable. If the enterprise cooperation is not fully aware of these risks, and in time control, enterprises may be in a very passive position, the cooperative innovation inputs outweigh the benefits, and even lead to the failure of the cooperation, resulting in enormous economic and social loss. Therefore, risk analysis of enterprise cooperative innovation has important theoretical and practical significance for the sustainable development of enterprises.

Enterprise cooperative technology innovation is a process from research and development, production to the market, so the risk prevention and control of enterprise cooperative innovation are necessary in order to combine the whole process of innovation. By looking at the risk of abnormal links, and then to adjust the significant risk, so that the risk status can quickly transfer from the serious risk to the lower risk. At present, the research on the risk of enterprise cooperative technology innovation has been gradually expanded in the field, but it still remains in a single subject. O’Keefe gives the applying principles of innovation to curriculum revision.⁴ Babalola et al. give the evaluation of factors influencing technological innovations of small and medium enterprises in Nigerian industrial estates.⁵ Zhang and Lv discuss the mediating role of supply chain learning in intellectual capital and technological innovation.⁶ Fulbright discusses the solving methodology by using inventive problem in incorporating innovation into iterative software development.⁷ Wang discusses the organizations and innovation within the biotechnology industry in China from an interdisciplinary perspective.⁸ Xiaoan studies the conflict in collaborative design built on the network environment. It is considered that coordination is a very important process to cope with the collision, and the method to deal with it is obtained.⁹ Li et al. present the risk evaluation of technology innovation in China’s oil and gas industry.¹⁰ Jian and Yudong research the virtual enterprise gains and risk distribution, risk management and supervision, and proposed a dynamic contract system based on risk transfer algorithm of risk control and optimization model and design the risk checklist.¹¹ Stefanie and Nancy contrast risk perceptions of technology-based service innovations in inter-organizational settings.¹² Greenstein and Feinman study e-commerce security and risk management, review the electronic commerce laws and regulations environment, payment mechanism, intelligent agent, marketing network and Internet encryption and verification and safety and risk management problems.¹³ Köhler and Som put forward the risk preventative innovation strategies for emerging technologies – the cases of nano-textiles and smart textiles.¹⁴

The complexity and uncertainty of the enterprise cooperation technology innovation means that the cooperative technology innovation has the characteristic of nonlinearity.^15,16 The nerve network is suitable for recognizing and simulating nonlinear system, and the wavelet transformation or the decomposition displays the good time frequency localization characteristic and the multi-criteria function; the wavelet nerve network based on the wavelet decomposition and the nerve network have the good fault-tolerant ability and the non-linearity approaching performance, and therefore, can be used to discuss the model of enterprise cooperation technology innovation.

In this paper, the wavelet analysis method is considered in order to construct the analytical framework for the risk prevention and control of enterprise cooperative technology innovation. Based on the control theory and risk theory, risk prevention and control method of enterprise cooperative technology innovation are constructed from the angle of risk state transfer and risk mechanism improvement.

Analysis of seeking the risk of collaborative innovation of enterprise

By means of seeking the risk unusual link, and then carrying on the adjustment to a remarkable risk, the risk condition may be transferred from the acute risk condition to the lower risk condition in a comparatively quick manner.^17,18 The risk gradually inspection method can be used for seeking the risk abnormal links, the steps are as follows:

A series of risk controlling points in the entire process should be established, just as the figure 1 showing.

The various risks controlling point Rt should be calculated.

And then the risk increment between the risk control points should be calculated, the maximal risk increment should be found by carrying on the comparison, therefore, the abnormal link is the innovation unit between the risk controlling point t and the risk controlling point t+1.

Generally speaking, cooperative technology innovation risk includes six aspects: IT risk, market risk, environmental risk, performance risk, capital risk and collaborative risk. Increasing the input of a link reduces the size of a certain risk, and hence there is a negative correlation between each risk and input. These relationships can be described by a negative Logistic curve, that is

r_{i} (t) = r_{i 0} (t) {1 - 1 / [1 + (1 - a_{i} (t)) \cdot e^{- b_{i} (t) \cdot (x_{i} (t) - x_{i 0} (t))}]}

where

i = 1, 2, \dots, 6

t = 1, 2, \dots, n

r_{i} (t)

is the risk size of

i

risk species when the risk control investment quantity increases to

x_{i} (t)

at the link

t

r_{i 0} (t)

is the risk size of the

i

risk species at the link

t

at the moment before risk control.

x_{i 0} (t)

and

x_{i} (t)

are the total quantity of risk control investment of the

i

risk species at the link

t

at the moment of before and after risk control. Different types of risks correspond to different parameters,

a_{i} (t), b_{i} (t) \in (0, 1)

; they are related to the reduction degree of

i

risk species. The bigger of , the greater of declining degree along with the increase in investment of No. risk, can be determined based on similar risk of historical data records by using statistical methods. Decision makers can use the above relationships to reduce the critical risk

r_{i_{0}} (t_{0})

in the risk exception, thereby reducing the total risk

R

Concept of controlling risk regulation gradient of collaborative innovation of enterprise

There does exist a complex non-linear function mechanism between the risk factors of enterprise collaborative innovation; hence we can use the non-linear method to research the keeping way and controlling mechanism of cooperative technology innovation of enterprise. The every abnormal link in figure 1 will be inspected from the system angle, and the investment plan of controlling risk of the link must be made for optimization and adjustment control through the whole process of cooperative technology innovation, the superior risk can be obtained, and the goal of controlling risk will be achieved, and achieves the goal of controlling risk. This kind of coordination control pursues the smallest controlling cost of the composite system, and gets the best controlling effect.^19–21 Therefore, the convenient and external expression method can be used to reflect the coordinated degree and can be considered as the basis for harmoniously controlling risk.

Figure 1.

Risk controlling point of collaborative innovation of enterprise.

And now, we define the risk regulation elasticity of the investment

E_{x_{i} (t)} = \frac{Δ r_{i} (t)}{r_{i} (t)} / \frac{Δ x_{i} (t)}{x_{i} (t)} = \frac{\partial r_{i} (t)}{\partial x_{i} (t)} \cdot \frac{x_{i} (t)}{r_{i} (t)}

(1)

(i = 1, 2, \dots, 6, t = 1, 2, \dots, n)

where

r_{i} (t)

is the risk size of

i

risk species when the risk control investment quantity increases to

x_{i} (t)

at the link

t

x_{i} (t)

is the total quantity of risk control investment of

i

risk species at the link

t

. And the relation between

r_{i} (t)

and

x_{i} (t)

is presented in equation (1).

And we define $c_{i} (t)$ that expresses the influencing coefficient of risk state variable $r_{i} (t)$ , $r_{i} (t)$ is the risk size of $i$ risk species at the link $t$ , and it can be obtained by the following equation

c_{i} (t) = \frac{a_{i} (t)}{\sum_{t = 1}^{n} \sum_{i = 1}^{6} a_{i} (t)} a_{i} (t) = \sum_{q \neq i} a_{i q}^{t} r_{q}

where

r_{q}

is the risk state variable of

q

a_{i q}^{t}

is the influencing factor of the risk state variable

r_{i} (t)

i

to influence the risk state variable

r_{q}

q

at the link

t

, and can be obtained from the historical data and experience of the experts and the related personnel who enjoy the collaborative innovation.

a_{i} (t)

is the influencing variable of the risk state variable

r_{i} (t)

i

risk species at the phase

t

. The coordination coefficient

c_{i} (t)

expresses the total influencing degree of

r_{i} (t)

to the other risk state variables, and reflects the influencing degree of

r_{i} (t)

, and embodies the basis for harmoniously controlling risk.

By the synthetical risk regulation elasticity and its influencing coefficient, we can obtain the controlling risk regulation gradient of collaborative innovation of enterprise

H_{i} (t) = c_{i} (t) \cdot E_{x_{i} (t)} (i = 1, 2, \dots, 6, t = 1, 2, \dots, n)

Thus, the risk regulation matrix $H (t)$ of collaborative innovation of enterprise is the basis for harmoniously controlling risk that includes all the links between the head and tail risk control point

H (t) = [\begin{array}{l} H_{1} (1) & H_{2} (1) & \dots & H_{6} (1) \\ H_{1} (2) & H_{2} (2) & \dots & H_{6} (2) \\ \dots & \dots & \dots & \dots \\ H_{1} (n) & H_{2} (n) & \dots & H_{6} (n) \end{array}]

If the risk regulation investment total quantity is $x$ , the risk regulation investment quantity $x_{i} (t)$ of $i$ risk species at the link $t$ is as follows

\begin{array}{l} x_{i} (t) = \frac{H_{i} (t)}{\sum_{t = 1}^{n} \sum_{i = 1}^{6} H_{i} (t)} \cdot x \\ (i = 1, 2, \dots, 6, t = 1, 2, \dots, n) \end{array}

Model of coordinating and controlling risk of collaborative innovation of enterprise

Based on the above analysis and control theory viewpoint, we bring forward a complete controlling risk model of the collaborative innovation of enterprise, and this controlling risk model is based on the wavelet and nerve network, as shown in Figure 2.^22–24

Figure 2.

Sketch-map of the overall risk regulation.

Coordination coefficient $C = (c_{1} (1), c_{2} (1), \dots, c_{5} (n), c_{6} (n))$ , regulation elasticity $E = (E_{x_{1} (1)}, E_{x_{2} (1)}, \dots, E_{x_{5} (n)}, E_{x_{6} (n)})$ , coordination control vector $X = (x_{1} (1), x_{2} (1), \dots, x_{5} (n), x_{6} (n))$ , risk state vector $r =$ $(r_{1}, r_{2}, r_{3}, r_{4}, r_{5}, r_{6})$ , $r_{i}$ ( $i = 1, 2, \dots, 6$ ), respectively, express the size of environment risk, IT risk, market risk, fund risk, cooperation risk and performance risk; the total risk is $R = \sum_{i = 1}^{6} r_{i}$ , so the coordination control problem can be described, regarding the coordination matrix $H$ , defines the total risk expected value $R$ , choosing the wavelet nerve network algorithm, obtains the coordination optimization value of system control input $X$ by unceasingly optimization. The characteristic of regulative algorithm is to preparatory confirm the system input data by regulation degree, and the data will be calculated by the educated wavelet nerve network, that will greatly enhance the computation speed and accuracy.

In essence, wavelet transformation is a kind of integral transformation of different parameter spaces

W (\vec{p}) = \int_{- \infty}^{+ \infty} f (\vec{x}) h (\vec{p}, \vec{x}) d \vec{x}, f (\vec{x}) \in L^{2} (R^{n})

In the equation,

f (\vec{x})

is the function that has the right collection.

h (\vec{p}, \vec{x})

is the wavelet base.

\vec{p} = (P_{1}, P_{2}, …, P_{n})

is the coordinate of n dimensional parameter space.

\vec{x} = (x_{1}, x_{2}, …, x_{m})

is the coordinate of m dimensional parameter space.

Under the situation of the uni-dimensional signal $f (x)$ , the wavelet base $h (\vec{p}, \vec{x})$ and $h (a, b, x)$ (here, $\vec{p} = (a, b)$ ) may be deduced by the following equation

h (a, b, x) = \frac{1}{\sqrt{| a |}} h (\frac{x - b}{a})

In the equation,

h (x)

is the basic wavelet or the female wavelet,

{| a |}^{- 1 / 2}

is the normalized coefficient,

a, b

are, respectively, the contraction and translation factors of

h (a, b, x)

, and regarding the signal

f (x)

, the partial structure differentiation of the factor can be realized by adjusting parameters

a, b

and the window size and position of wavelet base. In the formula, h(x) is the basic wavelet.

The wavelet nerve network is the kind of front nerve network based on the wavelet analysis of the nerve network, and may be regarded as a new function of joining nerve network which takes the wavelet as the base.²⁵ For calculating the overall risk regulation input of collaborative innovation of enterprise, and if the total quantity of imputing the training sample is $P$ , the network output pitch point has six, the input pitch point has 6n, and then to the $p$ sample, the $i$ pitch point, the output may be indicated by the following equation

{\tilde{r}}_{i}^{p} = σ (u_{i}) = σ [\sum_{k = 1}^{K} ω_{i k} \sum_{m = 1}^{6 n} x_{m} h (\frac{x_{m} - b_{k}}{a_{k}})] (i = 1, 2, \dots, 6)

In the equation,

σ (x) = 1 / [1 + \exp (- x)]

is the sigmoid function. Figure 3 is the corresponding framework sketch-map of wavelet nerve network. And the significance of hiding level summation is that its signal indication will be realized by linearly summing the selecting wavelet base, the parameter of needing study has

ω_{i k}

b_{k}

a_{k}

(

i = 1, 2, \dots, 6

k = 1, 2, \dots, K

), and K is the wavelet element which need to select, and can be optimized by the following energy function with smallest mean error

E = \frac{1}{2} \sum_{p = 1}^{P} \sum_{i = 1}^{6} {(r_{i}^{p} - {\tilde{r}}_{i}^{p})}^{2}

Here,

r i_{}^{p}

is the ideal output of the

P

sample and the

i

pitch point.

Figure 3.

Framework sketch-map of the risk regulation wavelet nerve network.

There are various wavelet function, here the wavelet base is calculated by the following formula, namely²⁶

h (x) = \cos (1.75 x) \exp (- x^{2} / 2)

This wavelet is the Gauss wave of cosine modulation, the time domain function and the corresponding frequency range function resolution is higher.

If ${x^{'}}_{m} = (x_{m} - b_{k}) / a_{k}$ , $σ^{'} (u) = \partial σ (u) / \partial u = σ (u) [1 - σ (u)]$ , and the gradient of $E$ _, respectively, is $\frac{\partial E}{\partial ω_{i k}} = - \sum_{p = 1}^{P} \sum_{m = 1}^{6 n} (r_{i}^{p} - {\tilde{r}}_{i}^{p}) σ^{'} (u_{i}) x_{m} \cos (1.75 {x^{'}}_{m}) \exp (- {x^{'}}_{m}^{2} / 2)$

\begin{matrix} \frac{\partial E}{\partial b_{k}} = - \sum_{p = 1}^{P} \sum_{i = 1}^{6} \sum_{m = 1}^{6 n} (r_{i}^{p} - {\tilde{r}}_{i}^{p}) σ^{'} (u_{i}) x_{m} ω_{i k} [1.75 \sin (1.75 {x^{'}}_{m}) \\ \exp (- {x^{'}}_{m}^{2} / 2) + \cos (1.75 {x^{'}}_{m}) \exp (- {x^{'}}_{m}^{2} / 2) {x^{'}}_{m}] / a_{k} \end{matrix}

\begin{matrix} \frac{\partial E}{\partial a_{k}} = - \sum_{p = 1}^{P} \sum_{i = 1}^{6} \sum_{m = 1}^{6 n} (r_{i}^{p} - {\tilde{r}}_{i}^{p}) σ^{'} (u_{i}) x_{m} ω_{i k} [1.75 \sin (1.75 {x^{'}}_{m}) \\ \exp (- {x^{'}}_{m}^{2} / 2) + \cos (1.75 {x^{'}}_{m}) \exp (- {x^{'}}_{m}^{2} / 2) {x^{'}}_{m}] {x^{'}}_{m} / a_{k}^{2} \end{matrix}

And $i = 1, 2, \dots, 6$ , $k = 1, 2, \dots, K$ _.

Taking the conjugate grads method to optimize network parameter $ω_{i k}$ , $b_{k}$ , $a_{k}$ , and suppose $ω = (ω_{11}, ω_{12}, \dots, ω_{6 K})$ and $\partial E / \partial ω = (\partial E / \partial ω_{11}, \partial E / \partial ω_{12}, \dots, \partial E / \partial ω_{6 K})$ , and $s (ω^{j})$ is the circulation search vector of $j$ searching direction, and $ω^{j}$ is the connecting vectors from the hiding level to outputting level of $j$ for circulation search

s (ω^{j}) = {\begin{array}{l} - \frac{\partial E}{\partial ω^{j}} & j = 1 \\ - \frac{\partial E}{\partial ω^{j}} + \frac{- \frac{\partial E}{\partial ω^{j}} {[- \frac{\partial E}{\partial ω^{j}}]}^{T}}{- \frac{\partial E}{\partial ω^{j - 1}} {[- \frac{\partial E}{\partial ω^{j - 1}}]}^{T}} s (ω^{j - 1}) & j \neq 1 \end{array}

The same principle is applied to $s (b^{j})$ and $s (a^{j})$ . Therefore, the network parameter vector may be adapted by the following equation

ω^{j + 1} = ω^{j} + α_{ω j} s (ω^{j})

b^{j + 1} = b^{j} + α_{b j} s (b^{j})

a^{j + 1} = a^{j} + α_{a j} s (a^{j})

α_{ω j}

α_{b j}

and

α_{a j}

are the fixing values between them

(0, 1)

According to the model of wavelet nerve network analog and regulation matrix, we seek the natural law of regulating risk of collaborative innovation of enterprise. The regulation error is $e = \tilde{R} - R$ and $R$ is the expecting total risk. The regulation rate goal is $J = \min e^{2} / 2$ _.

The partial derivatives of system output to regulating input is $\partial R / \partial X = C \partial r_{i} / \partial X$ ( $i = 1, 2, \dots, 6$ ).

And $\partial R / \partial X = (\partial R / \partial x_{1}, \partial R / \partial x_{2}, \dots, \partial R / \partial x_{6 n})$ , $C = (1, 1, \dots, 1)$ .

\frac{\partial r}{\partial X} = [\begin{array}{l} \partial r_{1} / \partial x_{1} & \partial r_{1} / \partial x_{2} & \dots & \partial r_{1} / \partial x_{6 n} \\ \partial r_{2} / \partial x_{1} & \partial r_{2} / \partial x_{2} & \dots & \partial r_{2} / \partial x_{6 n} \\ \dots & \dots & \dots & \dots \\ \partial r_{6} / \partial x_{1} & \partial r_{6} / \partial x_{2} & \dots & \partial r_{6} / \partial x_{6 n} \end{array}]

$\frac{\partial r_{i}}{\partial x_{m}} = \sum_{k = 1}^{K} σ^{'} (u_{i}) ω_{i k} [\cos (1.75 {x^{'}}_{m}) \exp (- {x^{'}}_{m}^{2} / 2) - 1.75 x_{m} \sin (1.75 {x^{'}}_{m}) \exp (- {x^{'}}_{m}^{2} / 2) \cdot \frac{1}{a_{k}} - x_{m} \cos (1.75 {x^{'}}_{m}) \exp (- {x^{'}}_{m}^{2} / 2) {x^{'}}_{m} \cdot \frac{1}{a_{k}}]$ _. The process of regulating algorithm is as follows:

$j_{\max}$ is the biggest iteration number of times, $e_{\max}$ is the expecting precision.

$X^{j + 1} = X^{j} + α_{X j} C s (X^{j})$ and $0 < α_{X j} < 0$ .

\begin{array}{l} X^{0} = [\frac{H_{1} (1)}{\sum_{t = 1}^{n} \sum_{i = 1}^{6} H_{i} (t)} \cdot x, \frac{H_{2} (1)}{\sum_{t = 1}^{n} \sum_{i = 1}^{6} H_{i} (t)} \cdot \\ x, \dots, \frac{H_{6} (n)}{\sum_{t = 1}^{n} \sum_{i = 1}^{6} H_{i} (t)} \cdot x] \end{array}

s (X^{j}) = {\begin{array}{l} - \frac{\partial r_{i}}{\partial X^{j}} & j = 1 \\ - \frac{\partial r_{i}}{\partial X^{j}} + \frac{- \frac{\partial r_{i}}{\partial X^{j}} {[- \frac{\partial r_{i}}{\partial X^{j}}]}^{T}}{- \frac{\partial r_{i}}{\partial X^{j - 1}} {[- \frac{\partial r_{i}}{\partial X^{j - 1}}]}^{T}} s (X^{j - 1}) & j \neq 1 \end{array}

(3) If

j = j_{\max}

e_{j} \leq e_{\max}

, then stop the calculation, otherwise return to the previous steps.

Finally, $X$ is the optimized regulation input value of satisfying the total risk demand.

Illustration

X Company is ready to use cross cooperation innovation mode of research and development stage and industrialization stage. The research and development and production of the main parts of the product and supporting parts will be crossly carried out . The main parts of the product will be immediately produced after the completion of research and development, and the supporting parts are developed immediately at the same time. The supporting parts will be immediately produced after the completion of research and development. Thus, the lead time can be shortened and the matching degree and quality of the main parts and the supporting parts of the product can be improved.

As shown in Figure 4, the numbers 1, 2, 3, 4 and 5, respectively, represent the main parts of the product development, supporting parts of R&D, production of major parts of the product, production of the supporting parts and final product sales. c_11a, p_11a, c_12a, p_12a, c_21a, p_21a, c_22a, p_22a, c_3a, p_3a, respectively, represent the numerical risk and the probability risk of the above five links. By means of expert scoring method and analytic hierarchy process, the numerical risk and the probability risk are calculated. c_11a＝0.67, p_11a＝0.44, c_12a＝0.22, p_12a＝0.3, c_21a＝0.72, p_21a＝0.44, c_22a＝0.33, p_22a＝0.4, c_3a＝0.63, p_3a＝0.14

\begin{matrix} R_{1} = c_{11 a} p_{11 a} + c_{12 a} p_{12 a} (1 - p_{11 a}) + c_{21 a} p_{21 a} (1 - p_{11 a}) \\ + c_{22 a} p_{22 a} (1 - p_{1 a}) + c_{3} p_{3} (1 - p_{1 a}) (1 - p_{2 a}) = 0.45 \end{matrix}

Figure 4.

Interlinking of R&D stage and product stage.

The product innovation risk of X company is 0.45. When compared with the previous innovation risks, this is a general degree of risk. To fully protect the innovation success, the reasons for the risk should be eliminated, and make it smaller.To finding the risk of abnormal links, as shown in Figure 4, setting the end of four stages of 1, 2 and 3, 4, 5 as the risk control points, the risk increment of every risk control points $Δ_{t}$ is calculated as

\begin{matrix} Δ_{1} = c_{11 a} p_{11 a} = 0.2948; \\ Δ_{2, 3} = c_{12 a} p_{12 a} (1 - p_{11 a}) + c_{21 a} p_{21 a} (1 - p_{11 a}) = 0.1263 \\ Δ_{4} = c_{22 a} p_{22 a} (1 - p_{1 a}) = 0.0305 \\ Δ_{5} = c_{3} p_{3} (1 - p_{1 a}) (1 - p_{2 a}) = 0.0068 \end{matrix}

It can be seen that abnormal risk links are the first links, that is the main part research and development link of the product.

And then seeking optimal control input value. At first, establishing each link various risk control initial investment $x_{i} (t) = \frac{H_{i} (t)}{\sum_{t = 1}^{4} \sum_{i = 1}^{6} H_{i} (t)} \cdot x$ ( $i = 1, 2, \dots, 6$ , $t = 1, 2, \dots, 4$ ), $i$ is the risk types, mainly include environmental risk, IT risk, market risk, capital risk, collaborative risk and performance risk. is the risk control links, mainly include the main parts of the product development, the supporting parts R & D, the production of major parts of the product, the supporting parts production, and the final product sale. Based on historical data, using expert scoring, we can determine the risk control flexibility and its influence coefficient, thus get the regulatory matrix H(t), that is:

\begin{matrix} H (t) = [\begin{array}{l} H_{1} (1) & H_{2} (1) & H_{3} (1) H_{4} (1) H_{5} (1) & H_{6} (1) \\ H_{1} (2) & H_{2} (2) & H_{3} (2) H_{4} (2) H_{5} (2) & H_{6} (2) \\ H_{1} (3) & H_{2} (3) & H_{3} (3) H_{4} (3) H_{5} (3) & H_{6} (3) \\ H_{1} (4) & H_{2} (4) & H_{3} (4) H_{4} (4) H_{5} (4) & H_{6} (4) \end{array}] \\ = [\begin{array}{l} 0.4 & 0.7 & 0.2 & 0.8 & 0.9 & 0.7 \\ 0.5 & 0.6 & 0.4 & 0.6 & 0.4 & 0.6 \\ 0.1 & 0.1 & 0.4 & 0.04 & 0.6 & 0.8 \\ 0.8 & 0.5 & 0.6 & 0.5 & 0.6 & 0.4 \end{array}] \end{matrix}

According to the historical data, Managers of X company found that the total risk is 0.20, lower than the previous level, the estimated total initial investment is $x = 500$ million yuan. And the various types of risk control initial inputs is as follows

\begin{matrix} [\begin{matrix} x_{1} (1) & x_{2} (1) & x_{3} (1) x_{4} (1) x_{5} (1) & x_{6} (1) \\ x_{1} (2) & x_{2} (2) & x_{3} (2) x_{4} (2) x_{5} (2) & x_{6} (2) \\ x_{1} (3) & x_{2} (3) & x_{3} (3) x_{4} (3) x_{5} (3) & x_{6} (3) \\ x_{1} (4) & x_{2} (4) & x_{3} (4) x_{4} (4) x_{5} (4) & x_{6} (4) \end{matrix}] \\ = [\begin{array}{l} 0.4 & 0.7 & 0.2 & 0.8 & 0.9 & 0.7 \\ 0.5 & 0.6 & 0.4 & 0.6 & 0.4 & 0.6 \\ 0.1 & 0.1 & 0.4 & 0.04 & 0.6 & 0.8 \\ 0.8 & 0.5 & 0.6 & 0.5 & 0.6 & 0.4 \end{array}] \cdot \frac{1}{12.24} \times 500 \\ = [\begin{array}{l} 16.39 & 28.59 & 8.17 & 32.68 & 36.76 & 28.59 \\ 20.42 & 24.51 & 16.39 & 24.51 & 16.39 & 24.51 \\ 4.08 & 4.08 & 16.39 & 1.64 & 24.51 & 32.68 \\ 32.68 & 20.42 & 24.51 & 20.42 & 24.51 & 16.39 \end{array}] \end{matrix}

By means of Visual C++ programming software, firstly, the wavelet neural network is trained by related historical data of X company. For the results of innovation with satisfaction and dissatisfaction, two different types of four samples are selected, making a total of eight samples for learning. In order to speed up the network training, it is necessary to transform the original data of input and output to the interval $[0^{,} 1]$ and reload to the input and output layers of the network. According to the experience, 10 fixed wavelet bases will be adopted (that is and, taking some fixed values). The initial weight is a random number, the full adjustment optimization of weight, wavelet translation factor and dilation factor is carried out. The fixed values of $α_{ω 1}$ , $α_{b 1}$ and $α_{a 1}$ , and for 20 number training. For the innovative model to implement risk control, finally, the optimal control input value is 4.568 million yuan, which is lower than 5 million yuan. At this time, the regulation error is 0.0042.

Discussion and conclusion

The risk of each link in the enterprise cooperative technology innovation is not the same as the elasticity of the plan, that is, the amount of risk reduction obtained in the same situation is different. If the risk of an abnormal link is a flexible link, its regulation is very effective. Nevertheless, if its flexibility is small, even if the planned investment is very large, there is no guarantee that the risk has a significant reduction, so that the absolute risk is lowered to a satisfactory level. Therefore, in the case of inadequate investment plan, the plan should be to put into a larger flexibility in the link, so that the total risk of a significant reduction. For this reason, the effectiveness of the method is taken into consideration. And sometimes there is a need to consider the overall risk control. In this paper, we have put forward a method of coordinated control of enterprise cooperation technology innovation risk based on overall risk control. And then we give an example to illustrate the model and method.

There is existing complicated nonlinear interaction mechanism between the risk factors of enterprise cooperative technology innovation, and it is necessary to study the risk prevention and control mechanism of enterprise cooperative technology innovation. For each part of the figure 1, from the angle of investment, the risk control plan must be optimized and adjusted from different aspects in the whole risk control process, which will realize the global optimum of the overall risk, and achieve the purpose of risk control. This coordinated control pursuit complex system overall control of minimum cost, and the control effect is the best thing. Therefore, it is necessary to have an objective and convenient representation which not only reflects the degree of coordination of the composite system coordinated control but also can serve as a basis for the coordination and control. According to the principle of system science, this paper puts forward the concept of enterprise cooperative technology innovation risk control and then establishes the risk coordination control algorithm based on the theory of control.

In this article, we have assessed seeking the risk method of collaborative innovation of enterprises. Then the concept of controlling risk regulation gradient of collaborative innovation of an enterprise is presented. Finally, a complete controlling risk model of collaborative innovation of enterprise is advanced, and this controlling risk model is built on the wavelet and nerve network.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This paper was supported by National Natural Science Foundation of China (nos. U1404704 and 71702174), National Social Science Fund of China (no. 17BJY069).

References

Siyanbola

Olaopa

OR and

Hassan

OM.

Designing and implementing a science, technology and innovation policy in a developing country: recent experience from Nigeria. Ind Higher Educ 2013; 27: 323–331.

Zider

How venture capital works. Harv Bus Rev 1998; 6: 80–86.

Markowitz

HM.

Portfolio selection. J Finance 1952; 7: 77–91.

O’Keefe

Applying principles of innovation to curriculum revision. Int J Innov Sci 2013; 5: 173–178.

Babalola

Amiolemen

Adegbite

et al . Evaluation of factors influencing technological innovations of small and medium enterprises in Nigerian industrial estates. Int J Innov Sci 2015; 7: 39–54.

Zhang

H and

Intellectual capital and technological innovation: the mediating role of supply chain learning. Int J Innov Sci 2015; 7: 199–210.

Fulbright

Incorporating innovation into iterative software development using the inventive problem solving methodology. Int J Innov Sci 2013; 5: 203–212.

Wang

Organizations and innovation within the biotechnology industry in China: an interdisciplinary perspective. Int J Innov Sci 2014; 6: 167–176.

Xiaoan

Collaborative design method and technology based on the network environment and experimental study. Doctoral Dissertation, Chongqing University, China, 2001.

10.

Dong

Mikael

et al . Risk evaluation of technology innovation in China’s oil and gas industry. Int J Global Energy Issues 2013; 36: 1–12.

11.

Jian

Yudong

Construction and management of virtual enterprise. Beijing, China: Tsinghua University Publishing House, 2002.

12.

Stefanie

P and

Nancy

WV.

Contrasting risk perceptions of technology-based service innovations in inter-organizational settings. J Bus Res 2016; 69: 2424–2431.

13.

Greenstein

M and

Feinman

E-commerce security and risk management. Beijing, China: Huaxia Publishing House, 2001, p. 3.

14.

Köhler

AR and

Som

Risk preventative innovation strategies for emerging technologies the cases of nano-textiles and smart textiles. Technovation 2014; 34: 420–430.

15.

Clark-Wilson

Hoyles

C and

Noss

Scaling a technology-based innovation: windows on the evolution of mathematics teachers’ practices, ZDM. Int J Math Educ 2015; 47: 79–92.

16.

Kash

DE and

Rycoft

RW.

Patterns of innovating complex technologies: a framework for adaptive network strategies. Res Policy 2000; 29: 36–48.

17.

Shanley

CW.

Innovation and technology transfer: the challenge for large companies. Ind Higher Educ 2001; 15: 130–146.

18.

Zhu

Organisational culture and technology-enhanced innovation in higher education. Technol Pedagogy Educ 2015; 24: 65–79.

19.

Kogut

Joint ventures, the theoretical and empirical perspectives. Strateg Manag J 1988; 19: 319–332.

20.

Giambrone

Hemmings

Sturm

et al . Information technology innovation: the power and perils of big data. Br J Anaesth 2015; 115: 339–342.

21.

Bala

V and

Goyal

A noncooperative model of network formation. Econometrica 2000; 68: 61–68.

22.

Kim

Benchmarking government’s roles to assure the cooperation in collaborative technology innovation.

Benchmarking 2001; 8: 18–26.

23.

Chatterjee

Moody

Lowry

et al . Strategic relevance of organizational virtues enabled by informationtechnology in organizational innovation. J Manag Inf Syst 2015; 32: 158–196.

24.

Alvarez

LHR

Stenbacka

Adoption of uncertain multi-stage technology projects: a real options approach. J Math Econ 2001; 35: 86–93.

25.

Ramma

Samy

Gopee

Creativity and innovation in science and technology: bridging the gap between secondary and tertiary levels of education. Int J Educ Manage 2015; 29: 2–17.

26.

Ho D, Xu J and Zhou D. Wavelet neural networks for nonlinear system identification. In: Proceedings of the 39th IEEE conference on decision and control, Sydney, Australia, December 12-15, 2000, New York, USA: Institute of Electrical and Electronic Engineers, pp. 136–148.