Supply chain green emission reduction and coordination method based on Bayesian update inference

Abstract

The invention discloses a supply chain green emission reduction and coordination method based on Bayesian update inference, which is applied to a two-stage supply chain system composed of a product supplier and a product retailer, and is characterized in that the product retailer predicts the market demand of the next sales period according to historical data; the market demand prediction precision is improved through Bayesian update inference, then a retailer places an order to a supplier according to the predicted market demand after the precision is improved to report the order quantity, and the predicted market demand information is synchronously transmitted to the supplier to realize information sharing; through supply chain upstream and downstream collaborative integration and deployment, suppliers can adjust production tasks according to predicted market demands, it is guaranteed that participants of the whole supply chain can reasonably coordinate resources and adjust operation strategies under accurate market demand prediction, and green emission reduction and coordination of the supply chain are achieved. According to the method, the vicious circle of prediction distortion-excessive production-resource waste-carbon emission increase in a traditional supply chain is effectively broken.

Description

Technical Field

[0001] The present invention relates to the technical field of supply chain management, and particularly to a supply chain green emission reduction and coordination method based on Bayesian update inference. Background Art

[0002] With the increasing attention to the sustainable development of the supply chain, how to reduce carbon emissions while ensuring the operation efficiency of the supply chain has become a hot issue in academic and industrial research.

[0003] The traditional supply chain carbon emission management solutions have the following limitations:

[0004] 1. Low demand forecasting accuracy. Traditional demand forecasting methods usually perform simple statistical calculations based on historical data, such as moving average or exponential smoothing methods. These methods ignore the uncertainty and dynamic changes of market demand. Especially in the case of the gradual increase in consumers' environmental awareness and large fluctuations in the demand for green products, it is difficult to provide accurate demand forecasts, which easily leads to overproduction or insufficient inventory.

[0005] 2. Lack of effective information collaboration among supply chain members. In traditional supply chain management, the information barrier between upstream suppliers and downstream retailers leads to each member of the supply chain acting independently. The demand signals obtained by retailers are usually not shared with suppliers, and suppliers formulate production plans based on limited information, which may lead to supply-demand mismatches, resource waste, and an increase in carbon emissions.

[0006] 3. Lack of a systematic mechanism for carbon emission control. Existing supply chain carbon emission management methods mostly focus on end-of-pipe control, such as carbon tax policies and the guidance of consumers' environmental awareness. However, the potential for carbon emission control at the production end has not been fully exploited. Especially in terms of how supply chain members collaborate to reduce carbon emissions, there is still a lack of effective systematic solutions. Summary of the Invention

[0007] The present invention aims to avoid the deficiencies of the above-mentioned prior art, and provides a supply chain green emission reduction and coordination method based on Bayesian update inference to solve the problems of low demand forecasting accuracy, lack of effective information collaboration among supply chain members, lack of a systematic mechanism for carbon emission control in the existing supply chain carbon emission management solutions, and thus unsatisfactory carbon emission control.

[0008] The present invention adopts the following technical solutions to solve the technical problems:

[0009] The supply chain green emission reduction and coordination method based on Bayesian update inference of the present invention is applied to a two-level supply chain system composed of a product supplier and a product retailer. The product retailer predicts the market demand in the next sales period according to historical data, improves the accuracy of market demand prediction through Bayesian update inference, and then the retailer places an order with the supplier for the order quantity according to the predicted market demand after the accuracy improvement, and synchronously transmits the predicted market demand information to the supplier to achieve information sharing; through the collaborative integration and allocation of the upstream and downstream of the supply chain, the supplier can adjust the production tasks according to the predicted market demand, ensuring that all participants in the entire supply chain can reasonably coordinate resources and adjust business strategies under the accurate market demand prediction, so as to achieve green emission reduction and coordination of the supply chain.

[0010] The supply chain green emission reduction and coordination method based on Bayesian update inference of the present invention includes the following steps:

[0011] Step 1: The product retailer makes a preliminary prediction according to historical data to obtain the market demand prediction value D(p) in the next sales stage. The market demand prediction value D(p) is characterized by Equation (1):

[0012] D(p) = d(p) + ε (1)

[0013] Where:

[0014] d(p) is the mean value of the market demand, and d(p) = a - b×p;

[0015] a is the market size, that is, the potential demand for the product when the product price is 0;

[0016] b is the price elasticity coefficient, reflecting the sensitivity of consumers to price;

[0017] p is the retail price of the commodity;

[0018] ε is the demand prediction error, which is a random variable to reflect the uncertainty of demand. ε follows a prior normal distribution with a mean of 0 and a variance of ;

[0019] Step 2: The product retailer obtains the market demand expectation signal S in the next sales stage according to historical data and combined with future expectations. S = ε + ξ, where ξ is the demand expectation error, and ξ follows a normal distribution with a mean of 0 and a variance of σ 2 ;

[0020] According to the Bayesian method, the unbiased estimate of the demand prediction error ε is updated. The unbiased estimate E[ε|S] and the estimated variance VAR[ε|S] of the demand prediction error ε are obtained by Equation (2), and the posterior distribution of the demand prediction error ε is obtained.

[0021]

[0022] wherein

[0023] The market demand prediction expected value E[D(p)] is updated using the unbiased estimate E[ε|S], and E[D(p)] = d(p);

[0024] Furthermore, the updated value E1[D(p)] of the market demand prediction expectation is obtained as: E1[D(p)] = d(p) + S×ρ;

[0025] Based on the updated value E1[D(p)] of the market demand prediction expectation, the retail price p of the commodity and the order quantity q are determined, the order quantity q is transmitted to the supplier, and the market demand expectation signal S is transmitted to the supplier to achieve information sharing;

[0026] Step 3. Using the updated value E1[D(p)] of the market demand prediction expectation, the two - level supply chain system is simulated for real - world operation as follows:

[0027] Set the retail price decision target for the next sales stage of the product retailer as shown in Equation (3):

[0028]

[0029] w is the wholesale price of the supplier in the next sales stage;

[0030] e is the energy consumed per unit of commodity produced;

[0031] θ is the carbon emission per unit of energy;

[0032] t is the penalty cost borne by the retailer per unit of carbon emission;

[0033] By solving Equation (3), the retail price of the retailer in the next sales stage represented by the wholesale price w of the supplier in the next sales stage is obtained and the order quantity of the retailer in the next sales stage represented by the wholesale price w of the supplier in the next sales stage is obtained and are respectively characterized as: and

[0034] Set the optimal decision target of the supplier's wholesale price as shown in Equation (4):

[0035]

[0036] where c e is the energy consumption cost per unit of commodity;

[0037] Using the order quantity of the retailer in the next sales stage i.e., Q1(w), solve Equation (4) to obtain the wholesale price w1 of the supplier in the next sales stage;

[0038] Using the wholesale price \(w_1\) of the supplier in the next sales stage and the retail price \(P_1(w)\) and order quantity \(Q_1(w)\) of the retailer in the next sales stage represented by the wholesale price \(w\) of the supplier in the next sales stage, for the retail price of the retailer in the next sales stage represented by the wholesale price \(w\) of the supplier in the next sales stage and the sales volume of the retailer in the next sales stage represented by the wholesale price \(w\) of the supplier in the next sales stage are updated to obtain the retail price \(p\) and sales volume \(q\) of the retailer in the next sales stage.

[0039] The characteristics of the supply chain green emission reduction and coordination method based on Bayesian update inference of the present invention also lie in that when considering the improvement of the production technology of the supplier, setting the retail price decision target of the retailer in the next sales stage as formula (5):

[0040]

[0041] where \(x\) is the reduction in energy consumption per unit of commodity in the next sales stage;

[0042] Solve formula (5) to obtain the retail price of the retailer in the next sales stage represented by the wholesale price \(w\) of the supplier in the next sales stage and the reduction in energy consumption per unit of commodity \(x\) in the next sales stage and the order quantity of the retailer in the next sales stage represented by the wholesale price \(w\) of the supplier in the next sales stage and the reduction in energy consumption per unit of commodity \(x\) in the next sales stage and are respectively represented as: and

[0043] Set the decision target of the wholesale price \(w\) of the supplier in the next sales stage and the reduction in energy consumption per unit of commodity \(x\) in the next sales stage as formula (6):

[0044]

[0045] where:

[0046] C t is the production technology improvement cost, which refers to the technology improvement cost that the supplier needs to invest when the reduction in energy consumption per unit of commodity in the next sales stage is \(x\), C t =k×x 2 / 2, where \(k\) is a coefficient;

[0047] Using the order quantity of the retailer in the next sales stage i.e., \(Q_2(w,x)\), solve formula (6) to obtain the wholesale price \(w_2\) of the supplier in the next sales stage and the reduction in energy consumption per unit of commodity \(x_2\) of the supplier in the next sales stage;

[0048] Substitute the wholesale price \(w_2\) of the supplier in the next sales stage and the reduction in energy consumption per unit of commodity \(x_2\) of the supplier in the next sales stage into \(P_2(w,x)\) and \(Q_2(w,x)\) to obtain the retail price \(p\) and the sales volume \(q\) of the retailer in the next sales stage.

[0049] The characteristics of the supply chain green emission reduction and coordination method based on Bayesian update inference of the present invention also lie in:

[0050] When considering the environmental awareness of consumers, the market demand \(D(p)\) in the next stage is: \(D(p)=d(p)+\mu x+\varepsilon\)

[0051] where \(\mu>0\) characterizes the environmental awareness \(\mu\in(0,1]\) of consumers purchasing green products. The stronger the environmental awareness of consumers, the larger \(\mu\);

[0052] Set the decision-making goal of the retail price \(p\) of the retailer in the next sales stage as shown in Equation (7):

[0053]

[0054] Obtain the retail price of the retailer in the next sales stage expressed by the wholesale price \(w\) of the supplier and the reduction in energy consumption per unit of commodity \(x\) in the next sales stage through Equation (7) and obtain the order quantity of the retailer in the next sales stage expressed by the wholesale price \(w\) of the supplier and the reduction in energy consumption per unit of commodity \(x\) in the next sales stage and are respectively characterized as: and

[0055] Set the decision-making goals of the wholesale price \(w\) of the supplier in the next sales stage and the reduction in energy consumption per unit of commodity \(x\) in the next sales stage as shown in Equation (8):

[0056]

[0057] Utilize the order quantity of the retailer in the next sales stage i.e., \(Q_3(w,x)\), solve Equation (8) to obtain the wholesale price \(w_3\) of the supplier in the next sales stage and the reduction in energy consumption per unit of commodity \(x_3\) of the supplier in the next sales stage;

[0058] Substitute the wholesale price \(w_3\) of the supplier in the next sales stage and the reduction in energy consumption per unit of commodity \(x_3\) of the supplier in the next sales stage into \(P_3(w,x)\) and \(Q_3(w,x)\) to obtain the retail price \(p\) and the sales volume \(q\) of the retailer in the next sales stage.

[0059] Compared with the prior art, the beneficial effects of the present invention are reflected in:

[0060] The present invention constructs a demand forecasting model based on Bayesian updating. By analyzing historical sales data, market activity records, and real-time demand signals, retailers use Bayesian updating to dynamically adjust the posterior distribution of demand, effectively reducing the uncertainty of demand forecasting and significantly improving forecasting accuracy. Based on the information of accurate demand forecasting achieved by Bayesian updating, the coordination of the upstream and downstream of the supply chain is carried out, enabling both retailers and suppliers to reasonably allocate resources and adjust business strategies under the accurate forecast demand information, so as to achieve a reasonable balance between product supply and demand, thereby reducing the unnecessary carbon emission problems caused by the imbalance between supply and demand. Moreover, while reducing carbon emissions, the overall operating efficiency of the supply chain is significantly improved, providing an innovative solution for the sustainable development of the supply chain. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 FIG. is a comparison chart of overproduction carbon emissions under different signal noises when the method of the present invention is adopted or not, where is the over-carbon emission when the method of the present invention is not adopted, is the over-carbon emission when the method of the present invention is adopted;

[0062] Figure 2 FIG. is a comparison chart of the total carbon emissions of the supply chain under different cost efficiencies when the method of the present invention is adopted or not, where is the total carbon emission of the supply chain when the method of the present invention is not adopted, is the total carbon emission of the supply chain when the method of the present invention is adopted. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0063] In this embodiment, the supply chain green emission reduction and coordination method based on Bayesian updating inference is applied to a two-level supply chain system composed of a product supplier and a product retailer. The product retailer predicts the market demand in the next sales period according to historical data, improves the accuracy of market demand forecasting through Bayesian updating inference, and then the retailer places an order with the supplier for the order quantity according to the predicted market demand after the accuracy improvement, and synchronously transmits the predicted market demand information to the supplier to achieve information sharing; through the collaborative integration and allocation of the upstream and downstream of the supply chain, the supplier can adjust the production tasks according to the predicted market demand, ensuring that all parties involved in the entire supply chain can reasonably coordinate resources and adjust business strategies under the accurate market demand forecasting, and realizing the green emission reduction and coordination of the supply chain.

[0064] Specifically, it is carried out according to the following steps:

[0065] Step 1: The product retailer makes a preliminary prediction based on historical data to obtain the market demand prediction value D(p) in the next sales stage. The market demand prediction value D(p) is represented by Equation (1):

[0066] D(p) = d(p) + ε (1)

[0067] Wherein:

[0068] d(p) is the mean of the market demand, and d(p) = a - b×p;

[0069] a is the market size, that is, the potential demand for the product when the product price is 0;

[0070] b is the price elasticity coefficient, reflecting the sensitivity of consumers to price;

[0071] p is the retail price of the commodity;

[0072] ε is the demand forecasting error, which is a random variable to reflect the uncertainty of demand. ε follows a prior normal distribution with a mean of 0 and a variance of ;

[0073] Historical data includes the historical sales orders of the product, the sales records in promotional activities, and the online click volume of customers, etc.

[0074] Step 2: The product retailer obtains the market demand expectation signal S for the next sales stage based on historical data and combined with future expectations. S = ε + ξ, where ξ is the demand expectation error, that is, the random noise of the market demand expectation signal S observed by the retailer. For example, the extremely high promotional sales volume due to a large price discount, or the increase in demand due to a short-term price increase of competing products. ξ follows a normal distribution with a mean of 0 and a variance of σ 2 . When the variance of ξ decreases to 0, S = ε, which means that the retailer can obtain the exact distribution of demand from the signal S.

[0075] According to Bayes' method, the unbiased estimate of the demand forecasting error ε is updated. The unbiased estimate E[ε|S] and the estimated variance VAR[ε|S] of the demand forecasting error ε are obtained from Equation (2) to obtain the posterior distribution of the demand forecasting error ε;

[0076]

[0077] Where It must be less than 1. Therefore, the variance ρ·σ of the posterior distribution 2 is less than the variance of the prior distribution This means that the uncertainty of demand is reduced by information acquisition.

[0078] The unbiased estimate E[ε|S] is used to update the market demand forecasting expected value E[D(p)], and E[D(p)] = d(p);

[0079] Furthermore, the updated value E1[D(p)] of the market demand forecasting expectation is obtained as: E1[D(p)] = d(p) + S×ρ;

[0080] Determine the retail price p and order quantity q of the commodity according to the expected updated value E1[D(p)] of the market demand forecast, transmit the order quantity q to the supplier, and transmit the market demand expectation signal S to the supplier to achieve information sharing;

[0081] Step 3: Use the expected updated value E1[D(p)] of the market demand forecast to simulate the actual operation of the two-level supply chain system as follows. Starting from the retailer's decision, use the backward induction method to gradually iterate and deduce the decisions of the retailer and the supplier:

[0082] Set the retail price decision target of the product retailer in the next sales stage as shown in Equation (3):

[0083]

[0084] w is the wholesale price of the supplier in the next sales stage;

[0085] e is the energy consumed for producing a unit of commodity;

[0086] θ is the carbon emission generated by a unit of energy;

[0087] t is the penalty cost borne by the retailer for a unit of carbon emission;

[0088] By solving Equation (3), obtain the retail price of the retailer in the next sales stage expressed by the wholesale price w of the supplier in the next sales stage and obtain the order quantity of the retailer in the next sales stage expressed by the wholesale price w of the supplier in the next sales stage And respectively represent them as: and

[0089] Set the optimal decision target of the supplier's wholesale price as shown in Equation (4):

[0090]

[0091] where c e is the energy consumption cost per unit of commodity;

[0092] Use the order quantity of the retailer in the next sales stage i.e., Q1(w), to solve Equation (4) to obtain the wholesale price w1 of the supplier in the next sales stage;

[0093] Use the wholesale price w1 of the supplier in the next sales stage, and the retail price of the retailer in the next sales stage represented by the wholesale price w of the supplier in the next sales stage to represent P1(w) and the order quantity to represent Q1(w), for the retail price of the retailer in the next sales stage represented by the wholesale price w of the supplier in the next sales stage and the sales volume of the retailer in the next sales stage represented by the wholesale price w of the supplier in the next sales stage Update to obtain the retail price p and sales volume q of the retailer in the next sales stage.

[0094] When specifically implementing and considering the improvement of the supplier's production technology, set the decision-making goal of the retailer's retail price in the next sales stage as shown in Equation (5):

[0095]

[0096] where x is the reduction in energy consumption per unit of commodity in the next sales stage;

[0097] Solve Equation (5) to obtain the retailer's retail price in the next sales stage expressed by the supplier's wholesale price w in the next sales stage and the reduction in energy consumption per unit of commodity x in the next sales stage and the retailer's order quantity in the next sales stage expressed by the supplier's wholesale price w in the next sales stage and the reduction in energy consumption per unit of commodity x in the next sales stage and are respectively characterized as: and

[0098] Set the decision-making goals of the supplier's wholesale price w in the next sales stage and the reduction in energy consumption per unit of commodity x in the next sales stage as shown in Equation (6):

[0099]

[0100] where:

[0101] C t is the production technology improvement cost, which refers to the technology improvement cost that the supplier needs to invest when the reduction in energy consumption per unit of commodity in the next sales stage is x, C t = k×x 2 / 2, where k is a coefficient;

[0102] Using the retailer's order quantity in the next sales stage i.e., Q2(w,x), solve Equation (6) to obtain the supplier's wholesale price w2 in the next sales stage and the reduction in energy consumption per unit of commodity x2 of the supplier in the next sales stage;

[0103] Substitute the supplier's wholesale price w2 in the next sales stage and the reduction in energy consumption per unit of commodity x2 of the supplier in the next sales stage into P2(w,x) and Q2(w,x) to obtain the retailer's retail price p and sales volume q in the next sales stage.

[0104] When specifically implementing and considering the environmental awareness of consumers, the market demand D(p) in the next stage is:

[0105] D(p) = d(p) + μx + ε

[0106] Among them, μ > 0 characterizes the environmental awareness of consumers when purchasing green products, where μ ∈ (0, 1]. The stronger the environmental awareness of consumers, the larger μ is;

[0107] Set the decision-making goal of the retailer's retail price p in the next sales stage as shown in Equation (7):

[0108]

[0109] Obtain the retailer's retail price in the next sales stage expressed by the supplier's wholesale price w and the reduction in energy consumption per unit of goods x in the next sales stage through Equation (7) And obtain the retailer's order quantity in the next sales stage expressed by the supplier's wholesale price w and the reduction in energy consumption per unit of goods x in the next sales stage And are respectively characterized as: and

[0110] Set the decision-making goals of the supplier's wholesale price w in the next sales stage and the reduction in energy consumption per unit of goods x in the next sales stage as shown in Equation (8):

[0111]

[0112] Utilize the retailer's order quantity in the next sales stage That is, Q3(w, x), solve Equation (8) to obtain the supplier's wholesale price w3 in the next sales stage and the reduction in energy consumption per unit of goods x3 of the supplier in the next sales stage;

[0113] Substitute the supplier's wholesale price w3 in the next sales stage and the reduction in energy consumption per unit of goods x3 of the supplier in the next sales stage into P3(w, x) and Q3(w, x) to obtain the retailer's retail price p and sales volume q in the next sales stage.

[0114] If the supplier does not adopt the supply chain green emission reduction and coordination method of the present invention, but determines the wholesale price w according to the expected value E[D(p)] of the market demand forecast and the order quantity q of the supplier in the next sales stage, then the demand signal S is regarded as a normal random variable with a mean of 0 and a variance of of.

[0115] On this basis, to achieve maximum revenue, the retailer's retail price decision-making problem is:

[0116]

[0117] where w is the product wholesale price, and the unit carbon emission penalty cost borne by the retailer in the form of a carbon tax is e×θ×t.

[0118] where \(e\) is the energy consumption per unit of commodity produced, \(\theta\) is the carbon emission generated per unit of energy, which depicts the carbon emission generated per unit of energy consumption, and \(t\) is the penalty cost borne by the retailer per unit of carbon emission, that is, the penalty rate, which can measure the retailer's concern for environmental protection. The higher \(t\) is, the stronger the retailer's environmental concern is. For example, if the supplier produces \(Q\) units of products, it consumes \(Q\times e\) units of energy and generates \(Q\times e\times\theta\) carbon emissions.

[0119] When the supplier cannot obtain the accurate demand signal \(S\), the supplier can infer the distribution of demand information from the order quantity \(q\). In this process, the supplier can roughly estimate the possible general distribution and statistical indicators of the market demand, such as the mean and variance, etc. The supplier makes a decision on the wholesale price based on all the information it has. With the goal of maximizing the long-term expected profit, it uses the average order level, that is, the expected order quantity \(d(w + t\times e\times\theta) / 2\) as the demand for the next period to decide its wholesale price. This method of taking the mean is inaccurate as a prediction of future demand because the supplier does not know whether there was a promotion activity in the previous sales period, nor does it know the corresponding sales volume, and these activities play an irreplaceable role in affecting and predicting the real demand. Therefore, the supplier's decision problem is:

[0120]

[0121] When the supplier knows the demand signal \(S\) from the retailer, the supplier knows the exact value of the demand signal \(S\) when deciding the wholesale price \(w\), and uses the exact value of the order quantity as the demand estimate for the new sales period. Information sharing reduces the uncertainty of demand forecasting, and the supplier can directly regard the retailer's order quantity as the expected demand for the next month. The supplier's decision problem is:

[0122]

[0123] where \(c\) e is the cost per unit of energy consumption.

[0124] Solve for the order quantity, retail price, supplier's wholesale price, and overproduction carbon emission \(E\) w,C where \(E\) w,C =E[eθ·(q(S)-E[D])] + as shown in Table 1.

[0125] Table 1. Optimal decisions and carbon emissions of retailers and suppliers with and without information sharing before production technology improvement

[0126]

[0127] As can be seen from Table 1 above, under the coordinated forecasting and coordination method of the present invention, after the retailer shares the demand distribution updated by Bayesian with the supplier, the wholesale price and order quantity of the commodity are more matched with the actual demand, and the overproduction carbon emissions are reduced by 50%. This fully demonstrates the advantages and progressive significance of the present invention in reducing overproduction carbon emissions by constructing a demand forecasting model based on Bayesian update to allocate and coordinate the production and sales tasks of the entire supply chain compared with the traditional supply chain carbon emission management solution.

[0128] Figure 1 The figure below shows a comparison chart of overproduction carbon emissions under different signal noises when the method of the present invention is adopted or not. Among them is the over-carbon emission when the method of the present invention is not adopted, is the over-carbon emission when the method of the present invention is adopted. The overproduction carbon emission when the method of the present invention is adopted is half of that when the method of the present invention is not adopted. At the same time, the reduction of the variance of the demand expectation error ξ will increase the over-carbon emission, which indicates that when the demand expectation error is small, the method of the present invention can significantly reduce the overproduction carbon emission.

[0129] Assume that the unit energy consumption of the supplier is exogenous, and e can be regarded as the highest threshold of the unit energy consumption in the supplier's production process. In practice, many suppliers actively invest in improving green production technologies to save energy consumption. We incorporate the supplier's improvement of production technology to reduce energy consumption into the information sharing model. After the supplier improves the production technology, the decision-making problem of the retailer after observing the demand signal S is:

[0130]

[0131] where x is the reduction in unit commodity energy consumption.

[0132] When the supplier cannot obtain the accurate demand signal S, the supplier's decision-making problem is:

[0133]

[0134] When the retailer shares the signal S with the supplier, the supplier's decision-making problem is:

[0135]

[0136] C t is the production technology improvement cost, which refers to the technology improvement cost that the supplier needs to invest when the reduction in unit commodity energy consumption in the next sales stage is x. C t = k×x 2 / 2, where k is a coefficient.

[0137] Solve for the order quantity, retail price, supplier's wholesale price and order quantity, unit energy consumption reduction, and the total expected carbon emissions E of the supply chain in the cases where the retailer knows or does not know the demand signal S. C , where E C = E[e×θ×q(S)] and the carbon emissions from overproduction, as shown in Table 2 below:

[0138] Table 2. Optimal decisions and carbon emissions of retailers and suppliers with and without information sharing after production technology improvement

[0139]

[0140] From Table 2 above, it can be seen that under the prediction and coordination method of the present invention, after the retailer shares the Bayesian-updated demand distribution with the supplier, information sharing can significantly reduce the total carbon emissions of the supply chain. When the cost efficiency is low and b×(t×θ + c e ) 2 / 2 < k < 1, information sharing can reduce the carbon emissions from overproduction.

[0141] Figure 2 The figure below shows the comparison of the total carbon emissions of the supply chain under different cost efficiencies when the method of the present invention is or is not adopted, where is the total carbon emissions of the supply chain when the method of the present invention is not adopted, is the total carbon emissions of the supply chain when the method of the present invention is adopted. The total carbon emissions after adopting the present invention are always less than those without adopting the present invention, verifying that the present invention can reduce the total carbon emissions of the two-level supply chain system.

[0142] Under the same traditional attribute quality, green products often have a greater market demand than traditional products. We introduce consumers' environmental awareness into the demand function. When considering consumers' environmental awareness in purchasing green products, the demand D(p) = a - b×p + μ×x + ε at this time, where μ > 0 characterizes consumers' environmental awareness in purchasing green products, μ ∈ (0, 1], and the stronger the consumers' environmental awareness, the larger μ is. Consistent with consumers' preference for green products in reality, the overall demand for products increases linearly with the unit energy consumption reduction x because a higher x means a greater reduction in energy consumption and less carbon emissions in the production of unit products.

[0143] The retail price decision problem of the retailer is:

[0144]

[0145] When the supplier cannot obtain the accurate demand signal S, the supplier's decision problem is:

[0146]

[0147] When the retailer shares the signal S with the supplier, the supplier's decision-making problem is as follows:

[0148]

[0149] Solve for the order quantity, retail price, supplier's wholesale price and order quantity, unit energy consumption reduction, and the total expected carbon emissions E of the supply chain in the cases where the retailer knows or does not know the demand signal S. C , where E C = E[e×θ×q] and the carbon emissions from overproduction, as shown in Table 3:

[0150] Table 3. Optimal decisions and carbon emissions of retailers and suppliers with and without information sharing considering consumers' environmental awareness

[0151]

[0152] As can be seen from Table 3 above, information sharing reduces the total carbon emissions of the supply chain; if the environmental awareness coefficient μ of consumers satisfies then information sharing can reduce the carbon emissions from overproduction in the supply chain.

[0153] In summary, the present invention first enables the retailer to predict the market demand in the next sales period based on historical sales information, and then improves the accuracy of market demand prediction through Bayesian updating inference. The retailer then places an order with the supplier and reports the order quantity based on the predicted market demand information after the accuracy improvement and synchronizes the predicted market demand information to the supplier. Thus, the upstream and downstream of the supply chain are coordinated with each other based on this predicted information, enabling the supplier to adjust its production tasks according to the predicted market demand. Thereby, it is ensured that all parties involved in the entire supply chain can reasonably coordinate resources and adjust business strategies under accurate market demand prediction, so that the supply and demand of products are in order, thereby achieving the purpose of reducing carbon emissions in the supply chain.

[0154] Although the present invention has been described in detail above with general descriptions and specific embodiments, based on the present invention, some modifications or improvements can be made, which are obvious to those skilled in the art. Therefore, these modifications or improvements made without departing from the spirit of the present invention all fall within the scope of protection required by the present invention.

Claims

1. A supply chain green emission reduction and coordination method based on Bayesian update inference, which is applied to a two-level supply chain system composed of a product supplier and a product retailer, is characterized in that: The product retailer predicts the market demand in the next sales period based on historical data, improves the accuracy of market demand prediction through Bayesian update inference, and then the retailer places an order with the supplier according to the predicted market demand after the accuracy improvement and synchronously transmits the predicted market demand information to the supplier to achieve information sharing. Through the collaborative integration and allocation of the upstream and downstream of the supply chain, the supplier can adjust the production tasks according to the predicted market demand, ensuring that all participants in the entire supply chain can reasonably coordinate resources and adjust business strategies under the accurate market demand prediction, and achieving green emission reduction and coordination of the supply chain.

2. The supply chain green emission reduction and coordination method based on Bayesian update inference according to claim 1, characterized in that It includes the following steps: Step 1: The product retailer makes a preliminary prediction based on historical data to obtain the market demand prediction value D(p) in the next sales stage, and the market demand prediction value D(p) is characterized by Equation (1): D(p) = d(p) + ε (1) Where: d(p) is the mean value of the market demand quantity, and d(p) = a - b×p; a is the market size, that is, the potential demand quantity of the product when the product price is 0; b is the price elasticity coefficient, reflecting the sensitivity of consumers to price; p is the retail price of the commodity; ε is the demand forecasting error, which is a random variable to reflect the uncertainty of demand. ε follows a prior normal distribution with a mean of 0 and a variance of ; Step 2: The product retailer obtains the market demand expectation signal S for the next sales stage based on historical data and combined with future expectations, where S = ε + ξ, ξ is the demand expectation error, and ξ follows a normal distribution with a mean of 0 and a variance of σ 2 ; According to the Bayesian method, the unbiased estimate of the demand prediction error ε is updated, and the unbiased estimate E[ε|S] and the estimated variance VAR[ε|S] of the demand prediction error ε are obtained from Equation (2) to obtain the posterior distribution of the demand prediction error ε. Among them The unbiased estimate E[ε|S] is used to update the market demand prediction expected value E[D(p)], and E[D(p)] = d(p); Furthermore, the updated value E1[D(p)] of the market demand prediction expectation is: E1[D(p)] = d(p) + S×ρ; Based on the updated value E1[D(p)] of the market demand prediction expectation, the retail price p and the order quantity q of the commodity are determined, the order quantity q is transmitted to the supplier, and the market demand expectation signal S is transmitted to the supplier to achieve information sharing; Step 3: Using the updated value E1[D(p)] of the market demand prediction expectation, the simulation of the real operation of the two-level supply chain system is carried out as follows: Set the retail price decision target of the product retailer in the next sales stage as Equation (3): w is the wholesale price of the supplier in the next sales stage; e is the energy consumed per unit of commodity produced; θ is the carbon emission generated per unit of energy; t is the penalty cost borne by the retailer for unit carbon emission; By solving equation (3), the retailer's retail price in the next sales stage expressed by the supplier's wholesale price w in the next sales stage is obtained and the retailer's order quantity in the next sales stage expressed by the supplier's wholesale price w in the next sales stage is obtained and are respectively characterized as: and Set the optimal decision target of the supplier's wholesale price as Equation (4): Among them, c e is the energy consumption cost per unit commodity; Using the retailer's order quantity in the next sales stage That is, Q1(w), solve Equation (4) to obtain the supplier's wholesale price w1 in the next sales stage; Using the next sales stage supplier wholesale price w1 and the next sales stage retailer retail price representation P1(w) and order quantity representation Q1(w) represented by the next sales stage supplier wholesale price w, for the next sales stage retailer retail price represented by the next sales stage supplier wholesale price w and the next sales stage retailer sales volume represented by the next sales stage supplier wholesale price w are updated to obtain the next sales stage retailer retail price p and sales volume q.

3. The supply chain green emission reduction and coordination method based on Bayesian update inference according to claim 2, characterized in that: When considering the improvement of the supplier's production technology, set the retail price decision target of the product retailer in the next sales stage as Equation (5): Where, x is the reduction amount of unit commodity energy consumption in the next sales stage; Solve equation (5) to obtain the retailer's retail price in the next sales stage expressed by the supplier's wholesale price w in the next sales stage and the reduction in energy consumption per unit of goods x in the next sales stage as well as the retailer's order quantity in the next sales stage expressed by the supplier's wholesale price w in the next sales stage and the reduction in energy consumption per unit of goods x in the next sales stage and are respectively characterized as: and Set the decision targets of the wholesale price w of the supplier in the next sales stage and the reduction amount x of unit commodity energy consumption in the next sales stage as Equation (6): Where: C t The cost for production technology improvement refers to the technology improvement cost that the supplier needs to invest when the reduction in the energy consumption per unit of commodity in the next sales stage is x, and C t = k × x 2 / 2, where k is a coefficient; Using the retailer's order quantity in the next sales stage That is, Q2(w, x), solve equation (6) to obtain the supplier's wholesale price w2 in the next sales stage and the reduction in the supplier's energy consumption per unit of goods x2 in the next sales stage; Substitute the wholesale price w2 of the supplier in the next sales stage and the reduction amount x2 of unit commodity energy consumption of the supplier in the next sales stage into P2(w, x) and Q2(w, x) to obtain the retail price p and the sales volume q of the product retailer in the next sales stage.

4. The method for green emission reduction and coordination of the supply chain based on Bayesian update inference according to claim 2, characterized in that: When considering the environmental awareness of consumers, the market demand D(p) in the next stage is: D(p) = d(p) + μx + ε where μ > 0 characterizes the environmental awareness of consumers who purchase green products, μ ∈ (0, 1]. The stronger the environmental awareness of consumers, the larger μ is; Set the decision-making goal of the retailer's retail price p in the next sales stage as shown in Equation (7): Obtain the retailer's retail price in the next sales stage expressed by the supplier's wholesale price \(w\) and the reduction in energy consumption per unit of commodity \(x\) through equation (7). And obtain the retailer's order quantity in the next sales stage expressed by the supplier's wholesale price \(w\) and the reduction in energy consumption per unit of commodity \(x\). And respectively characterize them as: and Set the decision-making goals of the supplier's wholesale price w in the next sales stage and the reduction in energy consumption per unit of commodity x in the next sales stage as shown in Equation (8): Using the retailer's order quantity in the next sales stage That is, Q3(w,x), solve equation (8) to obtain the supplier's wholesale price w3 in the next sales stage and the reduction in the supplier's unit commodity energy consumption x3 in the next sales stage; Substitute the supplier's wholesale price w3 in the next sales stage and the reduction in energy consumption per unit of commodity x3 of the supplier in the next sales stage into the P3(w, x) and Q3(w, x), and obtain the retailer's retail price p and sales volume q in the next sales stage.

Images