A port collection and distribution planning method based on big data optimization technology

Abstract

The present invention discloses a port collection and distribution planning method based on big data optimization technology, comprising the following steps: using a collection and distribution cargo flow long-term prediction model and a collection and distribution cargo flow short-term prediction model to determine the possible value interval of the shipment flow, based on the possible value interval of the cargo flow, using the collection and distribution cost minimization principle in the planning area to construct a collection and distribution optimization model in the planning area, and constructing the constraint conditions of the collection and distribution optimization model in the planning area. The present invention solves the collection and distribution optimization model in the planning area based on the constraint conditions to obtain the planning result of the collection and distribution cargo flow distribution in the planning area, so that the planning result of the collection and distribution cargo flow distribution in the planning area obtained by solving takes into account both the stability of the cargo flow and the certainty of the cargo flow.

Description

Technical Field

[0001] The present invention relates to the technical field of logistics planning, and in particular to a port logistics planning method based on big data optimization technology. Background Art

[0002] As a transportation system for the concentration and evacuation of goods in ports, the port inland collection and distribution system is the link between ports, places of goods production and places of goods demand. It is the key to controlling the efficiency and benefits of port logistics services, and plays an important role in improving the efficiency of the entire container transportation system and reducing transportation costs. With the continuous development of the global container transportation industry, my country's container transportation industry also has new development opportunities. The container throughput of various ports has increased significantly, and the inland container collection and distribution system is also facing higher challenges. It is inevitable that the container collection and distribution network must be planned and constructed according to the needs of the modern integrated logistics industry, on the one hand to enhance its transportation capacity, and on the other hand to expand its service network. Therefore, it is of great significance to study how to scientifically plan the inland collection and distribution network and effectively invest in it.

[0003] In the existing technology, the planning of port logistics and distribution is optimized by multi-objective optimization of cargo flow distribution planning based on transportation cost and transportation efficiency during the planning process. However, the freight demand generated by inland cities of container ports is affected by multiple factors such as industry market, hinterland economy, and port facility level, and has obvious uncertainty in different time periods. Therefore, multi-objective optimization has not studied the cargo flow distribution of inland logistics and distribution systems under the condition of uncertain demand, resulting in insufficient system accuracy and system stability of port logistics and distribution planning. Summary of the invention

[0004] The purpose of the present invention is to provide a port logistics planning method based on big data optimization technology to solve the technical problem that the multi-objective optimization in the prior art does not study the cargo flow distribution of inland logistics system under uncertain demand conditions, resulting in insufficient system accuracy and system stability of port logistics planning.

[0005] In order to solve the above technical problems, the present invention specifically provides the following technical solutions:

[0006] A port collection and distribution planning method based on big data optimization technology includes the following steps:

[0007] Step S1, obtaining historical data of freight collection and distribution in the planning area, performing long-term deterministic prediction training and short-term stability prediction training based on the historical data of freight collection and distribution in the planning area, and obtaining a long-term prediction model for freight collection and distribution and a short-term prediction model for freight collection and distribution;

[0008] Step S2: Determine the possible value range of the shipment flow rate by using the long-term forecast model of the collection and distribution cargo flow rate and the short-term forecast model of the collection and distribution cargo flow rate;

[0009] Step S3, based on the possible value range of cargo flow, using the principle of minimizing the collection and distribution cost in the planning area, constructing an optimization model for the collection and distribution in the planning area;

[0010] Step S4: construct constraint conditions for the collection and distribution optimization model in the planning area, and solve the collection and distribution optimization model in the planning area based on the constraint conditions to obtain the planning result of the collection and distribution cargo flow allocation in the planning area.

[0011] As a preferred solution of the present invention, the construction of the long-term forecast model for the cargo collection and distribution flow includes:

[0012] Select the decisive factors of the freight collection and distribution flow, and extract the historical time series corresponding to the historical data of the freight collection and distribution flow in the planning area;

[0013] Obtain historical data of decisive factors in historical time series, and make one-to-one correspondence between the historical data of decisive factors and the historical data of freight collection and distribution flow in the planning area according to the historical time series;

[0014] The historical data of the decisive factors of the same historical time series are used as the input items of the BP neural network, and the historical data of the freight collection and distribution flow in the planning area of ​​the same historical time series are used as the output items of the BP neural network. The BP neural network is used to learn and train the input items of the BP neural network and the output items of the BP neural network to obtain the long-term prediction model of the freight collection and distribution flow, so as to realize the long-term deterministic prediction of the freight collection and distribution flow in the planning area;

[0015] The functional expression of the long-term forecast model of the cargo collection and distribution flow is:

[0016] O t =BP([S L,t |1≤L≤m]);

[0017] In the formula, O t is the cargo flow of the t-th time series, S L,t is the data value of the Lth decisive factor at the tth time series, m is the total number of decisive factors, and BP is the BP neural network.

[0018] As a preferred solution of the present invention, the construction of the short-term forecasting model for cargo collection and distribution flow includes:

[0019] The historical data of the freight collection and distribution flow in the planning area of ​​the preceding historical time series is used as the input item of the LSTM neural network, and the historical data of the freight collection and distribution flow in the planning area of ​​the following historical time series is used as the output item of the LSTM neural network. The input item of the LSTM neural network and the output item of the LSTM neural network are trained by using the LSTM neural network to obtain the short-term prediction model of the freight collection and distribution flow, so as to realize the short-term stability prediction of the freight collection and distribution flow in the planning area;

[0020] The functional expression of the short-term forecasting model for the cargo collection and distribution flow is:

[0021] O new =LSTM(O old );

[0022] In the formula, O new is the post-historical time series of cargo collection and distribution flow, O old is the cargo collection and distribution flow of the previous historical time series, and LSTM is the LSTM neural network.

[0023] As a preferred solution of the present invention, the determination of the possible value range of the cargo flow rate includes:

[0024] Counting the data values ​​of the decisive factors at the time sequence to be planned, inputting the data values ​​of the decisive factors at the time sequence to be planned into the long-term forecast model of the freight flow of the collection and distribution transportation, and obtaining the deterministic forecast value of the freight flow at the time sequence to be planned by the long-term forecast model of the freight flow of the collection and distribution transportation;

[0025] Obtain the data value of the cargo flow rate at the preceding time sequence of the time sequence to be planned, input the data value of the cargo flow rate at the preceding time sequence of the time sequence to be planned into the short-term prediction model of the cargo flow rate, and obtain the cargo flow stability prediction value at the time sequence to be planned by the short-term prediction model of the cargo flow rate;

[0026] The deterministic forecast value of cargo flow and the forecast value of cargo flow stability are respectively used as the two end values ​​of the interval, and a continuous interval formed by the two end values ​​is used as the possible value interval of cargo flow.

[0027] As a preferred solution of the present invention, the construction of the collection and distribution optimization model in the planning area includes:

[0028] The collection and distribution cost in the planning area is quantified. The quantification function of the collection and distribution cost in the planning area is:

[0029] E=E1+E2+E3+E4+E5;

[0030]

[0031]

[0032]

[0033]

[0034]

[0035] In the formula, C ikw,t is the unit freight cost of choosing the wth mode of transportation from the i-th city in the planning area to the k-th inland transfer station at the t-th time sequence, C ijw,t is the unit freight cost of choosing the wth mode of transportation from the i-th city in the planning area to the j-th port at the t-th time sequence, C kjw,t The unit freight cost of the wth mode of transportation selected from the kth inland transfer station in the planning area to the jth port at the tth time sequence, CI k is the unit operating cost of the kth inland transfer station in the planning area, CP j is the unit operating cost of the jth port in the planning area, X ikw,t is the inland transfer volume of the wth mode of transportation selected from the i-th city in the planning area to the k-th inland transfer station at the t-th time sequence, X kjw,t Y is the volume of inland transit volume in the tth time series from the kth inland transit station in the planning area to the jth port using the wth mode of transportation, kjw,t The cargo flow generated by the inland transit city that selects the wth mode of transportation from the kth inland transit station to the jth port in the tth time-series planning area, Z ijw,t is the direct transport volume of the wth mode of transport from the ith city in the planning area to the jth port at the tth time sequence, a is the total number of cities in the planning area, b is the total number of inland transfer stations in the planning area, c is the total number of ports in the planning area, n is the total number of transport modes in the planning area, i, j, k and w are counting variables, E1 is the sum of the transport costs of various modes of transport from each city in the planning area to each inland transfer station at the tth time sequence, E2 is the sum of the transport costs of various modes of transport from each inland transfer station in the planning area to each port at the tth time sequence, E3 is the sum of the transport costs of various modes of transport from each city in the planning area to each port at the tth time sequence, E4 is the sum of the operating costs of each inland transfer station at the tth time sequence, and E5 is the sum of the operating costs of each port at the tth time sequence;

[0036] Minimizing the collection and distribution cost in the planning area is used as the collection and distribution optimization model in the planning area. The model expression of the collection and distribution optimization model in the planning area is:

[0037] F t =min(E);

[0038] Among them, F tis the function value of the collection and distribution optimization model in the t-th timing planning area, min is the minimization operator, and E is the collection and distribution cost in the t-th timing planning area.

[0039] As a preferred solution of the present invention, the constraint conditions for constructing the collection and distribution optimization model in the planning area include:

[0040] Based on the transportation capacity constraints of the planned area, as well as the operating capacity constraints of ports and inland transfer stations, the constraints of the optimization model of the planned area are constructed, among which:

[0041] The function expression of the constraint condition is:

[0042]

[0043]

[0044]

[0045]

[0046] X ikw,t ≤N ikw,t *M w ;

[0047] X kjw,t +Y kjw,t ≤N kjw,t *M w ;

[0048] Z ijw,t ≤N ijw,t *M w ;

[0049] Where i∈A, j∈C, k∈B, w∈W, X ikw,t ≥0,X kjw,t ≥0, Y kjw,t ≥0, Z ijw,t ≥0,O i,t is the freight flow of the ith city in the tth time-series planning area, M w N is the unit transport capacity allowed by the carrier in the wth mode of transport, ikw,t The number of vehicles that choose the wth mode of transportation from the i-th city in the planning area to the k-th inland transfer station at the t-th time sequence, N ijw,t The number of vehicles that choose the wth mode of transportation from the i-th city in the planning area to the j-th port at the t-th time sequence, N kjw,t The number of vehicles that choose the wth mode of transportation from the kth inland transfer station in the planning area to the jth port at the tth time sequence, D kis the operating capacity of the kth inland transfer station, D j is the operating capacity of the jth port.

[0050] As a preferred solution of the present invention, the method of solving the collection and distribution optimization model in the planning area based on the constraint conditions to obtain the planning result of the collection and distribution cargo flow allocation in the planning area includes:

[0051] The possible value intervals of cargo flow are combined into the constraint conditions to obtain the bidirectional constraint conditions, and the search algorithm is used to solve the collection and distribution optimization model in the planning area based on the bidirectional constraint conditions to obtain X ikw,t , X kjw,t , Y kjw,t and Z ijw,t As the planning result of the distribution of cargo flow in the planning area, the planning result of the distribution of cargo flow in the planning area obtained by solving the problem takes into account both the stability and certainty of cargo flow.

[0052] As a preferred solution of the present invention, the number of the decisive factors is greater than or equal to 1, and the decisive factors are obtained by screening all the influencing factors of the cargo collection and distribution flow through principal components.

[0053] As a preferred solution of the present invention, the output items of the long-term forecast model for the collection and distribution of freight flow and the short-term forecast model for the collection and distribution of freight flow are normalized.

[0054] As a preferred embodiment of the present invention, the transportation mode includes railway transportation, road transportation and water transportation.

[0055] Compared with the prior art, the present invention has the following beneficial effects:

[0056] The present invention utilizes a long-term forecast model for collection and distribution of freight flow and a short-term forecast model for collection and distribution of freight flow to determine the possible value range of the shipment flow, and based on the possible value range of the freight flow, utilizes the principle of minimizing the collection and distribution cost in the planning area to construct an optimization model for collection and distribution in the planning area, constructs constraint conditions for the optimization model for collection and distribution in the planning area, and solves the optimization model for collection and distribution in the planning area based on the constraint conditions to obtain a planning result for the distribution of collection and distribution freight flow in the planning area, so that the planning result for the distribution of collection and distribution freight flow in the planning area obtained by the solution takes into account both the stability of the freight flow and the certainty of the freight flow. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the implementation methods of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for the implementation methods or the description of the prior art. Obviously, the drawings in the following description are only exemplary, and for ordinary technicians in this field, other implementation drawings can be derived from the provided drawings without creative work.

[0058] Figure 1 A flow chart of a port collection and distribution planning method provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0059] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0060] The freight demand generated by inland cities of container ports is affected by many factors such as industry market, hinterland economy, and port facility level, and has obvious uncertainty in different time periods. Therefore, multi-objective optimization has not studied the cargo flow distribution of the collection and distribution system under the condition of uncertain demand, resulting in insufficient system accuracy and system stability of port collection and distribution planning. Therefore, the present invention provides a method for the cargo flow distribution of the collection and distribution system under the condition of uncertain demand, and predicts the accuracy and stability of freight demand in a two-way manner, and solves the constructed collection and distribution optimization model using the two-way predicted freight demand (cargo flow), thereby obtaining the collection and distribution cargo flow distribution planning result that takes into account both cargo flow stability and cargo flow certainty, which can not only improve the overall level of the collection and distribution system, but also reduce the cost in the collection and distribution process, thereby effectively guiding the improvement of the port collection and distribution system.

[0061] like Figure 1 As shown, the present invention provides a port collection and distribution planning method based on big data optimization technology, comprising the following steps:

[0062] Step S1, obtaining historical data of freight collection and distribution in the planning area, performing long-term deterministic prediction training and short-term stability prediction training based on the historical data of freight collection and distribution in the planning area, and obtaining a long-term prediction model for freight collection and distribution and a short-term prediction model for freight collection and distribution;

[0063] Step S2: Determine the possible value range of the shipment flow rate by using the long-term forecast model of the collection and distribution cargo flow rate and the short-term forecast model of the collection and distribution cargo flow rate;

[0064] Step S3, based on the possible value range of cargo flow, using the principle of minimizing the cost of cargo collection and distribution in the planning area, construct an optimization model for cargo collection and distribution in the planning area;

[0065] Step S4: construct constraint conditions for the collection and distribution optimization model in the planning area, and solve the collection and distribution optimization model in the planning area based on the constraint conditions to obtain the planning result of the collection and distribution cargo flow allocation in the planning area.

[0066] The demand for logistics and distribution is determined by many factors, and the changes in various factors lead to the uncertainty of logistics and distribution demand. First, with the rapid development of science and technology in contemporary society, the product update cycle has become shorter, related competitive products have been continuously launched, and the number of commodities on the market has continued to increase, which has increased the difficulty of demand forecasting. At the same time, customers' order changes, order cancellations, and order additions will also cause changes in demand. Secondly, the turbulence of the trade market, the development of the urban economy, the fluctuation of transportation prices, the level of logistics and distribution services, and the adjustment of laws and regulations on container transportation will all cause fluctuations in logistics and distribution demand. In summary, under the joint influence of various factors, the demand for port logistics and distribution is volatile and cannot be regarded as a fixed value. Therefore, when optimizing the land logistics and distribution network, the uncertainty of port logistics and distribution demand should be fully considered to make the research results more realistic.

[0067] In summary, the present invention provides two cargo flow prediction models, one is a long-term prediction model for collection and distribution of cargo flow, and the other is a short-term prediction model for collection and distribution of cargo flow. Among them, the long-term prediction model for collection and distribution of cargo flow is a collection and distribution of cargo flow prediction model that is directly learned and trained based on the decisive factors of the collection and distribution of cargo flow. Therefore, it remains effective throughout the entire collection and distribution process and is a long-term effective prediction model. Moreover, the decisive factors will have a decisive influence on the collection and distribution of cargo flow. The collection and distribution of cargo flow predicted by the decisive factors is a causal determination model, which conforms to the accuracy estimation method of the collection and distribution of cargo flow, and embodies the long-term deterministic characteristics of the collection and distribution of cargo flow prediction.

[0068] The short-term forecasting model for the collection and distribution of freight flow is a model that predicts the collection and distribution of freight flow in the subsequent time series based on the collection and distribution of freight flow in the previous time series. It relies on the transient stability formed by the correlation between the previous and subsequent time series. For example, the turbulence of the trade market, the development of the urban economy, the fluctuation of transportation prices, the level of collection and distribution services, and the adjustment of laws and regulations on container transportation are small or equal to zero between short time intervals, resulting in small fluctuations in the collection and distribution demand (cargo flow). Therefore, it remains effective between adjacent time series in the entire collection and distribution process. It is a short-term effective forecasting model. The use of transient stability between time series to predict the collection and distribution of freight flow reflects the short-term stability characteristics of the collection and distribution of freight flow prediction.

[0069] The construction of the long-term forecast model for cargo flow in the collection and distribution industry includes:

[0070] Select the decisive factors of the freight collection and distribution flow, and extract the historical time series corresponding to the historical data of the freight collection and distribution flow in the planning area;

[0071] Obtain historical data of decisive factors in historical time series, and make one-to-one correspondence between the historical data of decisive factors and the historical data of freight collection and distribution flow in the planning area according to the historical time series;

[0072] The historical data of the decisive factors of the same historical time series are used as the input of the BP neural network, and the historical data of the freight collection and distribution flow in the planning area of ​​the same historical time series are used as the output of the BP neural network. The BP neural network is used to train the input and output of the BP neural network to obtain a long-term prediction model for the freight collection and distribution flow, so as to achieve long-term deterministic prediction of the freight collection and distribution flow in the planning area.

[0073] The functional expression of the long-term forecast model of cargo collection and distribution flow is:

[0074] O t =BP([S L,t |1≤L≤m]);

[0075] In the formula, O t is the cargo flow of the t-th time series, S L,t is the data value of the Lth decisive factor at the tth time series, m is the total number of decisive factors, and BP is the BP neural network.

[0076] The construction of the short-term forecasting model for cargo collection and distribution includes:

[0077] The historical data of the freight collection and distribution flow in the planning area of ​​the previous historical time series is used as the input of the LSTM neural network, and the historical data of the freight collection and distribution flow in the planning area of ​​the subsequent historical time series is used as the output of the LSTM neural network. The LSTM neural network is used to train the input and output of the LSTM neural network to obtain a short-term prediction model for the freight collection and distribution flow, so as to realize the short-term stability prediction of the freight collection and distribution flow in the planning area.

[0078] The functional expression of the short-term forecasting model for cargo collection and distribution is:

[0079] O new =LSTM(O old );

[0080] In the formula, O new is the post-historical time series of cargo collection and distribution flow, O old is the cargo collection and distribution flow of the previous historical time series, and LSTM is the LSTM neural network.

[0081] The determination of the possible value range of cargo flow includes:

[0082] Counting the data values ​​of the decisive factors at the time sequence to be planned, inputting the data values ​​of the decisive factors at the time sequence to be planned into the long-term forecast model of the freight flow of the collection and distribution transportation, and obtaining the deterministic forecast value of the freight flow at the time sequence to be planned by the long-term forecast model of the freight flow of the collection and distribution transportation;

[0083] Obtain the data value of the cargo flow rate at the preceding time sequence of the time sequence to be planned, input the data value of the cargo flow rate at the preceding time sequence of the time sequence to be planned into the short-term prediction model of the cargo flow rate, and obtain the cargo flow stability prediction value at the time sequence to be planned by the short-term prediction model of the cargo flow rate;

[0084] The deterministic forecast value of cargo flow and the stability forecast value of cargo flow are respectively used as the two end values ​​of the interval, and a continuous interval formed by the two end values ​​is used as the possible value interval of cargo flow.

[0085] Construct a possible value interval of cargo flow, use the long-term prediction model of cargo flow and the short-term prediction model of cargo flow to obtain the deterministic prediction value of cargo flow and the stability prediction value of cargo flow as the two interval endpoints of the possible value interval, thereby characterizing that one end of the interval has a long-term deterministic attribute and the other end has a short-term stability attribute. That is, taking the value towards the end of the interval with the long-term deterministic attribute indicates that the cargo flow prediction value has a stronger prediction accuracy, and the port collection and distribution planning based on the cargo flow prediction value will have a stronger time series accuracy, but there will be large adjustments and changes in the collection and distribution system between adjacent time series, resulting in instability of the collection and distribution system. Taking the value towards the end of the interval with the short-term stability attribute indicates that the cargo flow prediction value has a stronger prediction stability, and the port collection and distribution planning based on the cargo flow prediction value will have a stronger time series stability, but there will be a problem of insufficient accuracy of the collection and distribution system in a single time series, resulting in insufficient planning accuracy of the collection and distribution system.

[0086] In summary, the present invention uses the deterministic prediction value of cargo flow and the stability prediction value of cargo flow as the two interval endpoint values ​​of the possible value range to construct the possible value range of cargo flow, which can take into account both long-term certainty and short-term stability in the possible value range of cargo flow, and ensure that the subsequent planning results solved according to the possible value range of cargo flow can take into account both the accuracy and stability of the collection and distribution system, and the collection and distribution system includes cities, ports, inland transfer stations, and transportation vehicles.

[0087] The construction of the collection and distribution optimization model in the planning area includes:

[0088] Quantify the collection and distribution costs in the planning area. The quantitative function of the collection and distribution costs in the planning area is:

[0089] E=E1+E2+E3+E4+E5;

[0090]

[0091]

[0092]

[0093]

[0094]

[0095] In the formula, C ikw,t is the unit freight cost of choosing the wth mode of transportation from the i-th city in the planning area to the k-th inland transfer station at the t-th time sequence, C ijw,t is the unit freight cost of choosing the wth mode of transportation from the i-th city in the planning area to the j-th port at the t-th time sequence, C kjw,t The unit freight cost of the wth mode of transportation selected from the kth inland transfer station in the planning area to the jth port at the tth time sequence, CI k is the unit operating cost of the kth inland transfer station in the planning area, CP j is the unit operating cost of the jth port in the planning area, X ikw,t is the inland transfer volume of the wth mode of transportation selected from the i-th city in the planning area to the k-th inland transfer station at the t-th time sequence, X kjw,t Y is the volume of inland transit volume in the tth time series from the kth inland transit station in the planning area to the jth port using the wth mode of transportation, kjw,t The cargo flow generated by the inland transit city that selects the wth mode of transportation from the kth inland transit station to the jth port in the tth time-series planning area, Z ijw,t is the direct transport volume of the wth mode of transport from the ith city in the planning area to the jth port at the tth time sequence, a is the total number of cities in the planning area, b is the total number of inland transfer stations in the planning area, c is the total number of ports in the planning area, n is the total number of transport modes in the planning area, i, j, k and w are counting variables, E1 is the sum of the transport costs of various modes of transport from each city in the planning area to each inland transfer station at the tth time sequence, E2 is the sum of the transport costs of various modes of transport from each inland transfer station in the planning area to each port at the tth time sequence, E3 is the sum of the transport costs of various modes of transport from each city in the planning area to each port at the tth time sequence, E4 is the sum of the operating costs of each inland transfer station at the tth time sequence, and E5 is the sum of the operating costs of each port at the tth time sequence;

[0096] Minimizing the collection and distribution cost in the planning area is used as the collection and distribution optimization model in the planning area. The model expression of the collection and distribution optimization model in the planning area is:

[0097] F t =min(E);

[0098] Among them, F t is the function value of the collection and distribution optimization model in the t-th timing planning area, min is the minimization operator, and E is the collection and distribution cost in the t-th timing planning area.

[0099] The constraints for constructing the collection and distribution optimization model in the planning area include:

[0100] Based on the transportation capacity constraints of the planned area, as well as the operating capacity constraints of ports and inland transfer stations, the constraints of the optimization model of the planned area are constructed, among which:

[0101] The function expression of the constraint condition is:

[0102]

[0103]

[0104]

[0105]

[0106] X ikw,t ≤N ikw,t *M w ;

[0107] X kjw,t +Y kjw,t ≤N kjw,t *M w ;

[0108] Z ijw,t ≤N ijw,t *M w ;

[0109] Where i∈A, j∈C, k∈B, w∈W, X ikw,t ≥0,X kjw,t ≥0, Y kjw,t ≥0, Z ijw,t ≥0,O i,t is the freight flow of the ith city in the tth time-series planning area, M w N is the unit transport capacity allowed by the carrier in the wth mode of transport, ikw,t The number of vehicles that choose the wth mode of transportation from the i-th city in the planning area to the k-th inland transfer station at the t-th time sequence, N ijw,t The number of vehicles that choose the wth mode of transportation from the i-th city in the planning area to the j-th port at the t-th time sequence, N kjw,t The number of vehicles that choose the wth mode of transportation from the kth inland transfer station in the planning area to the jth port at the tth time sequence, D k is the operating capacity of the kth inland transfer station, D j is the operating capacity of the jth port.

[0110] Based on the constraints, the optimization model of the collection and distribution in the planning area is solved to obtain the planning results of the collection and distribution cargo flow allocation in the planning area, including:

[0111] The possible value interval of cargo flow is combined into the constraint condition to obtain the bidirectional constraint condition, and the search algorithm is used to solve the collection and distribution optimization model in the planning area based on the bidirectional constraint condition to obtain X ikw,t , X kjw,t , Y kjw,t and Z ijw,t As the planning result of the distribution of cargo flow in the planning area, the planning result of the distribution of cargo flow in the planning area obtained by solving the problem takes into account both the stability and certainty of cargo flow.

[0112] The present invention takes cost minimization as the principle and constructs bidirectional constraints based on the possible value range of cargo flow that takes into account system accuracy and system stability, so as to obtain a planning result for the distribution of cargo flow in the planning area that takes into account the stability and certainty of the cargo flow process while maintaining the minimum cost.

[0113] The number of decisive factors is greater than or equal to 1, and the decisive factors are obtained by screening the principal components of all factors affecting the collection and distribution of freight flow.

[0114] The output items of the long-term forecast model for freight collection and distribution and the short-term forecast model for freight collection and distribution are normalized.

[0115] The modes of transportation include rail transportation, road transportation and water transportation.

[0116] The present invention utilizes a long-term forecast model for collection and distribution of freight flow and a short-term forecast model for collection and distribution of freight flow to determine the possible value range of the shipment flow, and based on the possible value range of the freight flow, utilizes the principle of minimizing the collection and distribution cost in the planning area to construct an optimization model for collection and distribution in the planning area, constructs constraint conditions for the optimization model for collection and distribution in the planning area, and solves the optimization model for collection and distribution in the planning area based on the constraint conditions to obtain a planning result for the distribution of collection and distribution freight flow in the planning area, so that the planning result for the distribution of collection and distribution freight flow in the planning area obtained by the solution takes into account both the stability of the freight flow and the certainty of the freight flow.

[0117] The above embodiments are only exemplary embodiments of the present application and are not intended to limit the present application. The protection scope of the present application is defined by the claims. Those skilled in the art may make various modifications or equivalent substitutions to the present application within the essence and protection scope of the present application, and such modifications or equivalent substitutions shall also be deemed to fall within the protection scope of the present application.

Claims

1. A port collection and distribution planning method based on big data optimization technology, characterized by: The following steps are involved: Step S1, obtaining historical data of freight collection and distribution in the planning area, performing long-term deterministic prediction training and short-term stability prediction training based on the historical data of freight collection and distribution in the planning area, and obtaining a long-term prediction model for freight collection and distribution and a short-term prediction model for freight collection and distribution; Step S2: Determine the possible value range of the shipment flow rate by using the long-term forecast model of the collection and distribution cargo flow rate and the short-term forecast model of the collection and distribution cargo flow rate; Step S3, based on the possible value range of cargo flow, using the principle of minimizing the collection and distribution cost in the planning area, constructing an optimization model for the collection and distribution in the planning area; Step S4, constructing the constraint conditions of the collection and distribution optimization model in the planning area, solving the collection and distribution optimization model in the planning area based on the constraint conditions to obtain the planning result of the collection and distribution cargo flow allocation in the planning area; The determination of the possible value range of cargo flow includes: Counting the data values ​​of the decisive factors at the time sequence to be planned, inputting the data values ​​of the decisive factors at the time sequence to be planned into the long-term forecast model of the freight flow of the collection and distribution transportation, and obtaining the deterministic forecast value of the freight flow at the time sequence to be planned by the long-term forecast model of the freight flow of the collection and distribution transportation; Obtain the data value of the cargo flow rate at the preceding time sequence of the time sequence to be planned, input the data value of the cargo flow rate at the preceding time sequence of the time sequence to be planned into the short-term prediction model of the cargo flow rate, and obtain the cargo flow stability prediction value at the time sequence to be planned by the short-term prediction model of the cargo flow rate; The deterministic forecast value of cargo flow and the forecast value of cargo flow stability are respectively used as the two end values ​​of the interval, and the continuous interval formed by the two end values ​​is used as the possible value interval of cargo flow; The construction of the collection and distribution optimization model in the planning area includes: Quantify the collection and distribution costs in the planning area. The quantitative function of the collection and distribution costs in the planning area is: E=E1+E2+E3+E4+E5; In the formula, C ikw,t is the unit freight cost of choosing the wth mode of transportation from the i-th city in the planning area to the k-th inland transfer station at the t-th time sequence, C ijw,t is the unit freight cost of choosing the wth mode of transportation from the i-th city in the planning area to the j-th port at the t-th time sequence, C kjw,t The unit freight cost of the wth mode of transportation selected from the kth inland transfer station in the planning area to the jth port at the tth time sequence, CI k is the unit operating cost of the kth inland transfer station in the planning area, CP j is the unit operating cost of the jth port in the planning area, X ikw,t is the inland transfer volume of the wth mode of transportation selected from the i-th city in the planning area to the k-th inland transfer station at the t-th time sequence, X kjw,t Y is the volume of inland transit volume in the tth time series from the kth inland transit station in the planning area to the jth port using the wth mode of transportation, kjw,t The cargo flow generated by the inland transit city that selects the wth mode of transportation from the kth inland transit station to the jth port in the tth time-series planning area, Z ijw,t is the direct transport volume of the wth mode of transport from the ith city in the planning area to the jth port at the tth time sequence, a is the total number of cities in the planning area, b is the total number of inland transfer stations in the planning area, c is the total number of ports in the planning area, n is the total number of transport modes in the planning area, i, j, k and w are counting variables, E1 is the sum of the transport costs of various modes of transport from each city in the planning area to each inland transfer station at the tth time sequence, E2 is the sum of the transport costs of various modes of transport from each inland transfer station in the planning area to each port at the tth time sequence, E3 is the sum of the transport costs of various modes of transport from each city in the planning area to each port at the tth time sequence, E4 is the sum of the operating costs of each inland transfer station at the tth time sequence, and E5 is the sum of the operating costs of each port at the tth time sequence; Minimizing the collection and distribution cost in the planning area is used as the collection and distribution optimization model in the planning area. The model expression of the collection and distribution optimization model in the planning area is: F t =min(E); Among them, F t is the function value of the collection and distribution optimization model in the t-th time-series planning area, min is the minimization operator, and E is the collection and distribution cost in the t-th time-series planning area; The constraints for constructing the collection and distribution optimization model in the planning area include: Based on the transportation capacity constraints of the planned area, as well as the operating capacity constraints of ports and inland transfer stations, the constraints of the optimization model of the planned area are constructed, among which: The function expression of the constraint condition is: X ikw,t ≤N ikw,t *M w ; X kjw,t +Y kjw,t ≤N kjw,t *M w ; From ijw,t ≤N ijw,t *M w ; Where i∈A, j∈C, k∈B, w∈W, X ikw,t ≥0,X kjw,t ≥0, Y kjw,t ≥0, Z ijw,t ≥0,O i,t is the freight flow of the ith city in the tth time-series planning area, M w is the unit transport capacity allowed by the carrier in the wth mode of transport, N ikw,t The number of vehicles that choose the wth mode of transportation from the i-th city in the planning area to the k-th inland transfer station at the t-th time sequence, N ijw,t The number of vehicles that choose the wth mode of transportation from the i-th city in the planning area to the j-th port at the t-th time sequence, N kjw,t The number of vehicles that choose the wth mode of transportation from the kth inland transfer station in the planning area to the jth port at the tth time sequence, D k is the operating capacity of the kth inland transfer station, D j is the operating capacity of the jth port.

2. According to claim 1, a port collection and distribution planning method based on big data optimization technology is characterized by: The construction of the long-term forecasting model for the cargo collection and distribution flow includes: Select the decisive factors of the freight collection and distribution flow, and extract the historical time series corresponding to the historical data of the freight collection and distribution flow in the planning area; Obtain historical data of decisive factors at the historical time series, and make one-to-one correspondence between the historical data of decisive factors and the historical data of cargo collection and distribution flow at the planning area according to the historical time series; The historical data of the decisive factors of the same historical time series are used as the input items of the BP neural network, and the historical data of the freight collection and distribution flow in the planning area of ​​the same historical time series are used as the output items of the BP neural network. The BP neural network is used to learn and train the input items of the BP neural network and the output items of the BP neural network to obtain the long-term prediction model of the freight collection and distribution flow, so as to realize the long-term deterministic prediction of the freight collection and distribution flow in the planning area; The functional expression of the long-term forecast model of the cargo collection and distribution flow is: O t =BP([S L,t |1≤L≤m]); In the formula, O t is the cargo flow of the t-th time series, S L,t is the data value of the Lth decisive factor at the tth time series, m is the total number of decisive factors, and BP is the BP neural network.

3. The port collection and distribution planning method based on big data optimization technology according to claim 1 is characterized by: The construction of the short-term forecasting model for cargo collection and distribution flow includes: The historical data of the freight collection and distribution flow in the planning area of ​​the preceding historical time series is used as the input item of the LSTM neural network, and the historical data of the freight collection and distribution flow in the planning area of ​​the following historical time series is used as the output item of the LSTM neural network. The input item of the LSTM neural network and the output item of the LSTM neural network are trained by using the LSTM neural network to obtain the short-term prediction model of the freight collection and distribution flow, so as to realize the short-term stability prediction of the freight collection and distribution flow in the planning area; The functional expression of the short-term forecasting model for the cargo collection and distribution flow is: O new =LSTM(O old ); In the formula, O new is the post-historical time series of cargo collection and distribution flow, O old is the cargo collection and distribution flow of the previous historical time series, and LSTM is the LSTM neural network.

4. According to the method of port collection and distribution planning based on big data optimization technology in claim 1, it is characterized in that: The planning result of the cargo flow distribution of the planned area is obtained by solving the optimization model of the planned area based on the constraint conditions, including: The possible value intervals of cargo flow are combined into the constraint conditions to obtain the bidirectional constraint conditions, and the search algorithm is used to solve the collection and distribution optimization model in the planning area based on the bidirectional constraint conditions to obtain X ikw,t , X kjw,t , Y kjw,t and Z ijw,t As the planning result of the distribution of cargo flow in the planning area, the planning result of the distribution of cargo flow in the planning area obtained by solving the problem takes into account both the stability and certainty of cargo flow.

5. According to claim 2, a port collection and distribution planning method based on big data optimization technology is characterized in that: The number of the decisive factors is greater than or equal to 1, and the decisive factors are obtained by screening all the influencing factors of the collection and distribution freight flow through the principal components.

6. The port collection and distribution planning method based on big data optimization technology according to claim 1 is characterized in that: The output items of the long-term forecast model for the collection and distribution of freight flow and the short-term forecast model for the collection and distribution of freight flow are normalized.

7. The port collection and distribution planning method based on big data optimization technology according to claim 1 is characterized in that: The transportation modes include railway transportation, road transportation and water transportation.

Images