A method for storing and scheduling a fracturing proppant

Abstract

The application discloses a fracturing proppant storage scheduling method, constructs a use amount prediction model, and predicts the use amount based on historical use records of the fracturing proppant, and provides the prediction result to a storage worker to assist the storage worker in replenishment, so that the requirement for the storage worker can be effectively reduced, and the use influencing factors of the fracturing proppant can be used as identification factors, so that more accurate demand prediction can be realized, meanwhile, real-time storage scheduling information can be received, and a storage robot can be controlled to intelligently schedule, so that the storage scheduling efficiency can be effectively improved, and compared with manual scheduling in the prior art, a large amount of time can be saved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of fracturing proppant scheduling, and particularly relates to a fracturing proppant storage scheduling method. BACKGROUND

[0002] Fracturing proppant is a kind of ceramic particle product with high fracturing strength, which is mainly used for downhole support in oil fields to increase oil and gas production and is an environmentally friendly product. As an effective way to transform oil and gas fields, fracturing technology has been widely known and applied. Fracturing technology is to use artificially applied super high pressure to cause different degrees of cracks in the rock formation, and then inject fluid into the cracks, so as to form a channel for easy oil and gas flow. In order to keep the cracks produced after fracturing in an open state, granular fracturing proppant is usually injected into the fluid.

[0003] In the storage scheduling of fracturing proppant, the goods are often taken by workers using forklifts and replenished by workers according to experience, so that the operation requirements are high and the taking efficiency is low. SUMMARY

[0004] The present application aims to provide a fracturing proppant storage scheduling method to solve the problems in the prior art.

[0005] The present application is implemented by the following technical solutions:

[0006] A fracturing proppant storage scheduling method comprises the following steps:

[0007] A neural network model is used to construct a use amount prediction model corresponding to the fracturing proppant, and a chaotic mapping mechanism is used to initialize the network parameters of the use amount prediction model to obtain network parameter individuals, wherein the network parameters include weight parameters and threshold parameters of the neural network model;

[0008] The fitness values corresponding to each network parameter individual are obtained, and the network parameter individual with the largest fitness value is taken as a leader, and the remaining network parameter individuals are divided into explorers and followers according to a preset proportion;

[0009] A neighborhood search algorithm is used to update the leader to obtain an updated leader, a cooperative search algorithm is used to update the explorer to obtain an updated explorer, and an adaptive search algorithm is used to update the follower based on the leader and the explorer to obtain an updated follower;

[0010] The target network parameter individual with the largest fitness value is selected from the updated leader, the updated explorer and the updated follower, and the target network parameter individual is taken as a global optimal value;

[0011] The updating of the leader, the explorer and the follower is repeated until the maximum number of training times is reached, the global optimal value is taken as the final network parameter of the use amount prediction model, and the trained use amount prediction model is obtained; in the training process, the leader, the explorer and the follower are re-divided every training period; each training period includes a plurality of training processes;

[0012] The historical use records of the pressure proppant corresponding to each target user are collected, and the trained use amount prediction model is used to identify the historical use records to obtain the predicted use amount corresponding to each target user. The predicted use amount corresponding to each target user is sent to the work terminal corresponding to the warehousing worker to assist the warehousing worker in restocking;

[0013] After sending the predicted use amount corresponding to each target user to the work terminal corresponding to the warehousing worker, real-time warehousing scheduling information transmitted by the work terminal is obtained, and the real-time warehousing scheduling information includes the type and quantity of the fracturing proppant;

[0014] According to the real-time warehousing scheduling information and the current warehousing information, the warehousing robot is planned to take the goods according to the predetermined goods taking route, and the warehousing robot is controlled to take the goods according to the predetermined goods taking route, so that the fracturing proppant storage scheduling is completed, and the warehousing efficiency is improved. The current warehousing information includes the type, quantity and shelf number of the warehousing fracturing proppant.

[0015] In a possible implementation, the fitness value corresponding to each network parameter individual is obtained, and the network parameter individual with the maximum fitness value is taken as the leader, and the remaining network parameter individuals are divided into explorers and followers according to a predetermined proportion, including:

[0016] The reciprocal of the error function of the use amount prediction model is taken as the fitness function, and the fitness value corresponding to each network parameter individual is obtained based on the fitness function;

[0017] The network parameter individual with the maximum fitness value is taken as the leader, M network parameter individuals in the remaining network parameter individuals are taken as the explorers, and N-1-M network parameter individuals are taken as the followers; wherein N represents the total number of network parameter individuals.

[0018] In a possible implementation, the leader is updated by using a neighborhood search algorithm to obtain an updated leader, including:

[0019] The historical optimal value of the leader in the historical training process and the global optimal value in the current updating process are obtained, and a first target vector corresponding to the historical optimal value of the leader minus the global optimal value in the current updating process is obtained;

[0020] Obtaining a norm of the first target vector, and attenuating the norm of the first target vector by a random number α generated according to a normal distribution N(0, 1) to obtain a first coefficient;

[0021] After multiplying the global optimal value in the current updating process by the first coefficient, a first updated value is obtained;

[0022] Adding the first updated value to the global optimal value in the current updating process to obtain an updated leader.

[0023] In a possible implementation, the explorers are updated by using a cooperative search algorithm to obtain updated explorers, comprising:

[0024] Obtaining historical optimal values of all explorers in a historical training process;

[0025] Randomly selecting one explorer from all explorers to obtain a target explorer;

[0026] Randomly selecting one explorer from all explorers except the target explorer to obtain a partner corresponding to the target explorer, and determining a historical optimal value corresponding to the partner;

[0027] Obtaining a mean vector between the historical optimal value of the target explorer and the historical optimal value corresponding to the partner;

[0028] Obtaining a step factor based on a current training number, and attenuating the step factor by a random number α generated according to a normal distribution N(0, 1) to obtain a second coefficient;

[0029] After multiplying the mean vector by the second coefficient, a second updated value is obtained, and the mean vector is added to the second updated value to obtain an updated target explorer;

[0030] Iterating all explorers to obtain updated explorers.

[0031] In a possible implementation, the leader and the explorers are used as a basis to update the followers by using an adaptive search algorithm to obtain updated followers, comprising:

[0032] Obtaining a reverse learning individual of the leader;

[0033] Determining a first decision interval (0, 0.33], a second decision interval (0.33, 0.66], and a third decision interval (0.66, 1), and randomly generating a decision factor between (0, 1) for each follower;

[0034] When the decision factor is located in the first decision interval, the follower is guided and updated based on the leader to obtain an updated follower;

[0035] When the decision factor is located in the second decision interval, a follower is randomly determined with an explorer, and the follower is guided and updated based on the determined explorer, to obtain an updated follower;

[0036] When the decision factor is located in the third decision interval, the follower is guided and updated based on the reverse learning individual, to obtain an updated follower.

[0037] In a possible implementation, when the decision factor is located in the first decision interval, the follower is guided and updated based on the leader, to obtain an updated follower as:

[0038] When the decision factor is located in the first decision interval, a second target vector is obtained by subtracting the historical optimal value of the follower from the global optimal value;

[0039] A norm of the second target vector is obtained, and a third coefficient is obtained by dividing the norm of the second target vector by an interval length of the network parameter; the interval length is used to represent a difference between an upper limit and a lower limit of the network parameter;

[0040] A fourth coefficient is determined based on the fitness of the follower in the previous two training processes;

[0041] A first update step factor of the follower is obtained according to the third coefficient and the fourth coefficient;

[0042] A fifth coefficient is obtained by attenuating the first update step factor of the follower with a random number α generated according to a normal distribution N(0, 1);

[0043] A third update value corresponding to the follower is obtained by multiplying the average of the global optimal value and the historical optimal value of the follower by the fifth coefficient;

[0044] The average of the global optimal value and the historical optimal value of the follower is added to the third update value, to obtain an updated follower.

[0045] In a possible implementation, when the decision factor is located in the second decision interval, a follower is randomly determined with an explorer, and the follower is guided and updated based on the determined explorer, to obtain an updated follower as:

[0046] When the decision factor is located in the second decision interval, a third target vector is obtained by subtracting the historical optimal value of the follower from the historical optimal value corresponding to the determined explorer;

[0047] A norm of the third target vector is obtained, and a sixth coefficient is obtained by dividing the norm of the third target vector by an interval length of the network parameter; the interval length is used to represent a difference between an upper limit and a lower limit of the network parameter;

[0048] The seventh coefficient is determined based on the follower's fitness during the previous two training processes;

[0049] Obtaining a second update step factor of the follower according to the sixth coefficient and the seventh coefficient;

[0050] The random number α generated by the normal distribution N(0,1) is used to attenuate the second update step factor of the follower to obtain the eighth coefficient;

[0051] Multiply the average of the determined historical optimal value corresponding to the explorer and the historical optimal value of the follower by the eighth coefficient to obtain the fourth updated value corresponding to the follower;

[0052] The average of the determined historical optimal value corresponding to the explorer and the historical optimal value of the follower is added to the fourth update value to obtain an updated follower.

[0053] In a possible implementation, when the decision factor is in the third decision interval, the follower is guided and updated based on the reverse learning individual, and the updated follower is:

[0054] When the decision factor is in the third decision interval, the fourth target vector is obtained by using the reverse learning individual minus the historical optimal value of the follower;

[0055] Obtaining a norm of the fourth target vector and dividing the norm of the fourth target vector by the interval length of the network parameter to obtain a ninth coefficient; wherein the interval length is used to represent the upper limit minus the lower limit of the network parameter;

[0056] Based on the fitness of the follower during the previous two training processes, the tenth coefficient is determined;

[0057] Obtaining a third update step factor of the follower according to the ninth coefficient and the tenth coefficient;

[0058] The third update step factor of the follower is attenuated using a random number α generated by the normal distribution N(0,1) to obtain the eleventh coefficient;

[0059] Multiply the average of the historical optimal values ​​of the reverse learning individual and the follower by the eleventh coefficient to obtain the fifth update value corresponding to the follower;

[0060] The average of the historical optimal values ​​of the reverse learning individual and the follower is added to the fifth updated value to obtain the updated follower.

[0061] In one possible implementation, a fracturing proppant storage scheduling process is performed by planning a pick-up route for a storage robot based on real-time storage scheduling information and current storage information, and controlling the storage robot to pick up the product along the predetermined pick-up route.

[0062] According to the real-time storage scheduling information and the current storage information, a pickup node to be traversed by the storage robot is determined;

[0063] Based on the pickup node to be traversed by the storage robot, a pickup path code is obtained; a first element in the pickup path code is a position of the storage robot, and a last element in the pickup path code is a delivery position;

[0064] A plurality of different pickup path codes are repeatedly obtained, and the plurality of pickup path codes are obtained;

[0065] A travel time corresponding to each pickup path code is obtained, and the travel time is used to represent a time required for the storage robot to advance at a uniform speed according to the pickup path code;

[0066] A pickup path code with a minimum travel time is determined from the plurality of pickup path codes, and a target pickup path code is obtained;

[0067] A mutation operation and a crossover operation are performed on the pickup path code, and an updated pickup path code is obtained;

[0068] The target pickup path code is re-determined from the original pickup path code and the updated pickup path code;

[0069] It is judged whether the travel time corresponding to the target pickup path code is not reduced for N rounds, if yes, the target pickup path code is output, and if not, a half code with a maximum travel time is eliminated from the original pickup path code and the updated pickup path code, and the step of performing the mutation operation and the crossover operation is returned;

[0070] According to the current storage information, the storage robot is controlled to pick up at each pickup node according to an order of the target pickup path code, and storage scheduling of the fracturing proppant is completed.

[0071] In a possible implementation, picking up at each pickup node includes:

[0072] RFID information or two-dimensional code information of the storage scheduling of the fracturing proppant is scanned, and current storage information of target storage fracturing proppant is obtained;

[0073] Based on the current storage information of the target storage fracturing proppant, the storage robot is controlled to pick up according to a requirement of the real-time storage scheduling information.

[0074] The application provides a fracturing proppant storage scheduling method, a use amount prediction model is constructed, the use amount can be predicted based on historical use records of the fracturing proppant, and the prediction result is provided to a storage worker to assist the storage worker in replenishment, the requirement for the storage worker can be effectively reduced, and the use influencing factors of the fracturing proppant can be used as identification factors, so that more accurate demand prediction is realized, meanwhile, real-time storage scheduling information can be received, and a storage robot is controlled to perform intelligent scheduling, the storage scheduling efficiency can be effectively improved, and compared with manual scheduling in the prior art, a large amount of time can be saved. BRIEF DESCRIPTION OF DRAWINGS

[0075] In order to more clearly illustrate the technical solutions of the example embodiments of the application, the drawings needed in the examples will be briefly introduced as follows, and it should be understood that the following drawings only show some of the embodiments of the application, and therefore should not be regarded as a limitation on the scope, and for those skilled in the art, other related drawings can also be obtained without creative labor.

[0076] Figure 1 A flowchart of a fracturing proppant storage scheduling method provided by the example of the application. DETAILED DESCRIPTION

[0077] In order to make the purpose, technical solutions and advantages of the application more clear and apparent, the application will be further described in detail below with reference to the examples and drawings, the illustrative embodiments of the application and the description thereof are only used to explain the application, and should not be regarded as a limitation on the application.

[0078] EXAMPLE

[0079] As shown in the drawings, the example of the application provides a fracturing proppant storage scheduling method, which comprises: Figure 1

[0080] S1, a neural network model is used to construct a use amount prediction model corresponding to the fracturing proppant, and a chaotic mapping mechanism is used to initialize network parameters of the use amount prediction model to obtain network parameter individuals, the network parameters include weight parameters and threshold parameters of the neural network model.

[0081] Optionally, the neural network model can be set as a convolutional neural network or a BP (Back Propagation, back propagation) neural network, and the example preferably uses a BP neural network as the use amount prediction model to identify the input matrix data or vector data, so as to obtain the predicted use amount.

[0082] ​In addition to using the chaotic mapping mechanism to initialize the network parameters of the usage prediction model, the network parameters can be randomly generated between the upper limit and the lower limit of the network parameters.

[0083] S2, obtain the fitness value corresponding to each network parameter individual, and take the network parameter individual with the maximum fitness value as the leader, and divide the remaining network parameter individuals into explorers and followers according to a preset proportion.

[0084] The total number L of input matrix data or vector data can be preset, the network parameter individual is applied to the usage prediction model, the actual output of the usage prediction model is obtained by taking the historical data of the previous L days as the input, and the error function value is obtained according to the actual output and the expected output. Taking the reciprocal of the error function value can obtain the fitness value. It should be noted that, in order to avoid the denominator of the fitness value being zero, a very small constant can be added, for example, the fitness value can be 1 / (error function value+0.00001).

[0085] Optionally, in addition to selecting the historical usage records of the proppant as the training data, the usage influencing factors of the proppant in the previous L days can also be used as the input data, and the historical data of the L+1 day is taken as the expected output, so as to further improve the usage prediction effect of the usage prediction model.

[0086] S3, the leader is updated by using a neighborhood search algorithm to obtain an updated leader. The explorer is updated by using a collaborative search algorithm to obtain an updated explorer. The follower is updated by using an adaptive search algorithm based on the leader and the explorer to obtain an updated follower.

[0087] The neighborhood search algorithm, the collaborative search algorithm and the adaptive search algorithm are used in this embodiment, which can effectively improve the training effect of the usage prediction model, so that the final predicted usage is more in line with the actual situation, and the requirement for the usage prediction model can be effectively reduced.

[0088] S4, the target network parameter individual with the maximum fitness value is selected from the updated leader, the updated explorer and the updated follower, and the target network parameter individual is taken as the global optimal value.

[0089] S5, the updating of the leader, the explorer and the follower is repeated until the maximum training number is reached, the global optimal value is taken as the final network parameter of the usage prediction model, and the trained usage prediction model is obtained. During the training process, the leader, the explorer and the follower are re-divided every training period. Each training period includes a plurality of training processes.

[0090] For example, the maximum number of training is 1000, and the leader, explorer and follower are re-divided every 10 times, but the historical optimal value of each individual still needs to be recorded for subsequent updating.

[0091] S6, collect the historical use record of the pressure proppant corresponding to each target user, and use the trained use amount prediction model to identify the historical use record to obtain the predicted use amount corresponding to each target user, and send the predicted use amount corresponding to each target user to the working terminal corresponding to the warehousing worker to assist the warehousing worker to replenish goods.

[0092] By predicting the use amount of the fracturing proppant, the experience requirement of the warehousing worker can be effectively reduced, thereby effectively assisting the warehousing worker to replenish goods.

[0093] S7, after sending the predicted use amount corresponding to each target user to the working terminal corresponding to the warehousing worker, acquiring real-time warehousing scheduling information transmitted by the working terminal, the real-time warehousing scheduling information including the type and quantity of the fracturing proppant.

[0094] Because the real-time warehousing scheduling information includes the type and quantity of the fracturing proppant, automatic distribution can be performed according to the storage information of various fracturing proppants in the current warehousing information. It is worth noting that the fracturing proppant forms a single packaged product when it is warehoused, and a two-dimensional code or RFID (Radio Frequency Identification) tag is provided on each product, and the type and quantity (or weight) of the single product are recorded in the two-dimensional code or RFID tag.

[0095] The current warehousing information can include the storage location, type and quantity (or weight) of each product, so that the goods can be taken from which storage points according to the total demand.

[0096] S8, according to the real-time warehousing scheduling information and the current warehousing information, planning the goods taking route of the warehousing robot, and controlling the warehousing robot to take goods according to the predetermined goods taking route, completing the storage scheduling of the fracturing proppant, and improving the warehousing efficiency. The current warehousing information includes the type, quantity and shelf number of the warehoused fracturing proppant.

[0097] The warehousing robot can be a single intelligent robot or a combination of multiple robots, which can travel along a specified route, scan a two-dimensional code or RFID tag, and then automatically take and deliver goods according to machine vision technology.

[0098] The fracturing proppant storage scheduling method provided by the application constructs a use amount prediction model, can predict the use amount based on the historical use records of the fracturing proppant, and provide the prediction result to the storage staff to assist the storage staff in restocking, can effectively reduce the requirements on the storage staff, and can also take the use influencing factors of the fracturing proppant as identification factors to realize more accurate demand prediction, can receive real-time storage scheduling information, and control the storage robot to intelligently schedule, can effectively improve the storage scheduling efficiency, and compared with the manual scheduling of the prior art, can save a large amount of time.

[0099] In a possible implementation, the fitness value corresponding to each network parameter individual is obtained, and the network parameter individual with the maximum fitness value is taken as the leader, and the remaining network parameter individuals are divided into explorers and followers according to a preset proportion, including:

[0100] The reciprocal of the error function of the use amount prediction model is taken as the fitness function, and the fitness value corresponding to each network parameter individual is obtained based on the fitness function.

[0101] The network parameter individual with the maximum fitness value is taken as the leader, M network parameter individuals in the remaining network parameter individuals are taken as the explorers, and N-1-M network parameter individuals are taken as the followers. N represents the total number of network parameter individuals.

[0102] In a possible implementation, the leader is updated by using a neighborhood search algorithm to obtain an updated leader, including:

[0103] The historical optimal value of the leader i in the historical training process is obtained and the global optimal value in the current updating process is obtained and the historical optimal value corresponding to the leader is obtained The first target vector is obtained by subtracting the global optimal value in the current updating process

[0104] In the first updating process, the leader is the global optimal value, but as the updating proceeds, a better individual than the leader may be generated, in order to ensure the overall training effect, the leader is always updated near the global optimal value.

[0105] The norm of the first target vector is obtained and the norm of the first target vector is attenuated by using a random number alpha generated according to the normal distribution N(0, 1) to obtain a first coefficient

[0106] The global optimal value in the current updating process is multiplied by the first coefficient ​​Afterwards, the first updated value is obtained

[0107]

[0108] The first updated value and the global optimal value in the current update process Add them together to get the updated leader

[0109] In one possible implementation, a collaborative search algorithm is used to update the explorer to obtain an updated explorer, including:

[0110] Get the historical optimal values ​​of all explorers during historical training Where j represents the explorer, j = 1, 2, …, J, J represents the total number of explorers, p represents the historical optimal value, and t represents the current number of training times.

[0111] Randomly select an explorer from all explorers and get the target explorer

[0112] In addition to the target explorer Randomly select an explorer from all explorers except Corresponding collaborators And determine the historical optimal value corresponding to the collaborator

[0113] Get target explorer The historical best value The historical optimal value corresponding to the collaborator The mean between , get the mean vector

[0114] Based on the current number of training times, obtain the step factor a1×(x max -x min )×exp(-30t / T) 10 , and use the random number α generated by the normal distribution N(0,1) to decay the step factor a1×(x max -x min )×exp(-30t / T) 10 , we get the second coefficient α×a1×(x max -x min )×exp(-30t / T) 10 Among them, a1 represents a random number between (0,1), x max Indicates the maximum value of the network parameter, x min represents the minimum value of the network parameter, and T represents the maximum number of training times.

[0115] After multiplying the mean vector by the second coefficient, a second updated value is obtained and adding the mean vector to the second updated value, an updated target explorer is obtained It is worth noting that the embodiment adopts direct multiplication of the mean vector and (x max -x min ) and in this case, the upper and lower limits of each dimension are the same, so direct multiplication is possible. If the upper and lower limits of a certain dimension are different, each dimension needs to be extracted and updated according to the formula shown in the updated target explorer. In the BP neural network, the weight and threshold are both between 0 and 1, so the embodiment does not consider the case where the upper and lower limits of the dimensions are different.

[0116] All explorers are traversed to obtain updated explorers.

[0117] In a possible implementation, the follower is updated based on the leader and the explorer to obtain an updated follower by using an adaptive search algorithm, and the method comprises the following steps:

[0118] An inverse learning individual of the leader is obtained x dmax represents the upper limit of the dth dimension parameter, x dmin represents the lower limit of the dth dimension parameter, represents the dth dimension parameter in the leader i, d = 1, 2, …, D, and D represents the total number of network dimensions, represents the dth dimension parameter in the inverse learning individual.

[0119] A first decision interval (0, 0.33], a second decision interval (0.33, 0.66], and a third decision interval (0.66, 1) are determined, and a decision factor Q between 0 and 1 is randomly generated for each follower. It is worth noting that the decision intervals can also be divided into different lengths.

[0120] When the decision factor Q is located in the first decision interval, the follower is guided and updated based on the leader to obtain an updated follower.

[0121] When the decision factor Q is located in the second decision interval, a certain explorer is randomly determined for the follower, and the follower is guided and updated based on the determined explorer to obtain an updated follower.

[0122] When the decision factor Q is located in the third decision interval, the follower is guided and updated based on the inverse learning individual to obtain an updated follower.

[0123] In a possible implementation, when the decision factor Q is in the first decision interval, the leader-based guided update is performed on the follower to obtain an updated follower as:

[0124] When the decision factor Q is in the first decision interval, the global optimal value is subtracted from the historical optimal value of the follower to obtain a second target vector wherein, represents the historical optimal value of the nth follower in the tth training process.

[0125] The norm of the second target vector is obtained The norm of the second target vector is used divided by the interval length x of the network parameter max -x min to obtain a third coefficient wherein, the interval length is used to represent the upper limit of the network parameter minus the lower limit.

[0126] The fourth coefficient is determined based on the fitness of the follower in the previous two training processes.

[0127] Optionally, the method for determining the fourth coefficient can include: in each round of update process, obtaining the fitness values of the follower after three kinds of update.

[0128] In the tth round of training process, for each kind of update, the fitness value F1 in the t-1th round of training process and the fitness value F2 in the t-2th round of training process are obtained.

[0129] According to the fitness value F1 and the fitness value F2 corresponding to each kind of update, the fitness difference values corresponding to the three kinds of update are ζ1, ζ2 and ζ3 respectively. Wherein, ζ1 represents the fitness difference value corresponding to the leader-based guided update, ζ2 represents the fitness difference value corresponding to the determined explorer-based guided update, and ζ3 represents the fitness difference value corresponding to the reverse learning individual-based guided update.

[0130] Therefore, according to ζ1, ζ2 and ζ3, the fourth coefficient φ can be obtained as: It is worth noting that according to the above formula, each follower corresponds to a fourth coefficient.

[0131] According to the third coefficient and the fourth coefficient φ, the first update step factor of the follower is obtained wherein, φ max represents the largest fourth coefficient among all followers, and N represents the total number of followers.

[0132] ​The random number α generated according to the normal distribution N(0,1) is used to attenuate the first update step factor β1 of the follower to obtain the fifth coefficient α×β1.

[0133] The global optimal value The historical best value with followers Average value Multiply by the fifth coefficient to get the third update value corresponding to the follower

[0134] Add the average of the global optimal value and the follower's historical optimal value to the third updated value to obtain the updated follower

[0135]

[0136] In a possible implementation, when the decision factor Q is in the second decision interval, an explorer is randomly determined for the follower, and based on the determined explorer, the follower is guided and updated, and the updated follower is obtained as follows:

[0137] When the decision factor Q is in the second decision interval, the historical optimal value corresponding to the determined explorer is used Subtract the follower's historical best value Get the third target vector

[0138] Get the norm of the third target vector And take the norm of the third target vector Divide by the interval length x of the network parameter max -x min , and get the sixth coefficient The interval length is used to represent the upper limit minus the lower limit of the network parameter.

[0139] The seventh coefficient is determined based on the fitness of the follower during the previous two training processes. The seventh coefficient is the same as the fourth coefficient and will not be repeated here.

[0140] According to the sixth coefficient and the seventh coefficient, the second update step factor of the follower is obtained.

[0141]

[0142] The random number α generated according to the normal distribution N(0,1) is used to attenuate the second update step factor of the follower to obtain the eighth coefficient α×β2.

[0143] The average of the historical optimal value corresponding to the determined explorer and the historical optimal value of the follower Multiply by the eighth coefficient α×β2 to get the fourth update value corresponding to the follower

[0144] The average of the historical optimal value of the explorer corresponding to the determined historical optimal value and the historical optimal value of the follower is added to the fourth update value to obtain an updated follower

[0145] In a possible implementation, when the decision factor Q is located in the third decision interval, the follower is guided to update based on the reverse learning individual, and an updated follower is obtained as follows:

[0146] When the decision factor Q is located in the third decision interval, the reverse learning individual is adopted Subtract the historical optimal value of the follower Obtain a fourth target vector

[0147] Obtain the norm of the fourth target vector And adopt the norm of the fourth target vector Divide by the interval length x of the network parameter max -x min , to obtain a ninth coefficient Wherein, the interval length is used to represent the upper limit of the network parameter minus the lower limit.

[0148] The tenth coefficient is determined based on the fitness of the follower in the previous two training processes. The tenth coefficient is the same as the fourth coefficient, which will not be described here.

[0149] According to the ninth coefficient and the tenth coefficient, a third update step factor of the follower is obtained

[0150]

[0151] The third update step factor of the follower is attenuated by a random number α generated according to the normal distribution N(0, 1) to obtain an eleventh coefficient α × β3.

[0152] The average of the reverse learning individual and the historical optimal value of the follower Multiplied by the eleventh coefficient α × β3, to obtain a fifth update value corresponding to the follower

[0153] The average of the reverse learning individual and the historical optimal value of the follower is added to the fifth update value to obtain an updated follower

[0154] Through the algorithm described in the embodiment, on the basis of ensuring the training effect, through cooperative search and self-adaptation, the global search and the local search can be well balanced, and the convergence speed of the network can be accelerated.

[0155] In a possible implementation, a picking route of the warehouse robot is planned according to real-time warehouse scheduling information and current warehouse information, and the warehouse robot is controlled to pick goods according to the established picking route to complete the fracturing proppant storage scheduling, including:

[0156] According to the real-time warehouse scheduling information and the current warehouse information, a picking node required to be traversed by the warehouse robot is determined.

[0157] For example, the current warehouse scheduling information requires B units of weight of A type products, the current warehouse information includes the position, type and weight of each product, and it is assumed that the packaging weight of each product is B / 5. Then, 5 A type products and their positions can be found according to the current warehouse information to meet the warehouse scheduling requirement. The found positions are the picking nodes required to be reached.

[0158] It is worth noting that only integer quantity input is accepted, or the weight of each product is an integer multiple.

[0159] A picking path code is obtained based on the picking node required to be traversed by the warehouse robot. The first element in the picking path code is the position of the warehouse robot, and the last element in the picking path code is the delivery position.

[0160] A plurality of different picking path codes are repeatedly obtained to obtain a plurality of picking path codes. For example, the starting point can be represented by 0, the ending point can be represented by 1, and other picking nodes are sequentially assigned with values (assuming that there are 5 nodes, and then 2, 3, 4, 5 and 6). At this time, the picking path code can be 0234561 or 0245361, so that the travel path can be obtained.

[0161] A travel time corresponding to each picking path code is obtained, and the travel time is used to represent the time required for the warehouse robot to advance at a constant speed according to the picking path code.

[0162] A picking path code with the minimum travel time is determined from the plurality of picking path codes to obtain a target picking path code.

[0163] A mutation operation and a crossover operation are performed on the picking path code to obtain an updated picking path code.

[0164] The mutation operation and the crossover operation are conventional operations of the genetic algorithm, and the data processing process after the mutation operation and the crossover operation is mainly described. After the mutation operation and the crossover operation, there can be the same values, at this time, the priority of the mutation position and the crossover position is improved, the mutation position and the crossover position are reserved, and the positions of other nodes are modified. For example, 0234561, after mutation, 0335561 is obtained, it can be found that the second position is changed from 2 to 3, and the fourth position is changed from 4 to 5, then the data processing is performed in the forward order, the second position is the mutation position and is reserved as 3, the value of the third position is the same as that of the second position and has not been mutated, and should be changed to 2, and the subsequent is the same. The processing can also be performed in the backward order, according to the reverse order, it can be found that the value of the third position before and after mutation is 5, and the fourth position is changed from 4 to 5, then the value of the fourth position is reserved, and the third position is changed to 4.

[0165] In summary, the other positions same as the mutation position are changed to other values, and it is guaranteed that the values of each pickup node appear.

[0166] The original pickup path code and the updated pickup path code are used to determine the target pickup path code.

[0167] It is judged whether the driving time corresponding to the target pickup path code is not reduced for N rounds, if yes, the target pickup path code is output, otherwise, half of the codes with the maximum driving time are eliminated from the original pickup path code and the updated pickup path code, and the step of performing the mutation operation and the crossover operation is returned.

[0168] According to the current storage information, the warehouse robot is controlled to pick up at each pickup node according to the order of the target pickup path code, and the fracturing proppant storage scheduling is completed.

[0169] In a possible implementation, picking up at each pickup node includes:

[0170] The RFID information or the two-dimensional code information of the fracturing proppant storage scheduling is scanned to obtain the current storage information of the target warehouse fracturing proppant.

[0171] The current storage information of the target warehouse fracturing proppant is used as a basis to control the warehouse robot to pick up according to the demand of the real-time storage scheduling information.

[0172] The above detailed description of the specific embodiments of the present application has been given to understand the purpose, technical solutions and beneficial effects of the present application. It should be understood that the above description is only a specific embodiment of the present application and is not used to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method of proppant storage scheduling, the method comprising: The application comprises the following steps: A neural network model is used to construct a fracturing proppant usage prediction model, and a chaotic mapping mechanism is used to initialize the network parameters of the usage prediction model to obtain network parameter individuals, which include weight parameters and threshold parameters of the neural network model; The fitness value corresponding to each network parameter individual is obtained, and the network parameter individual with the maximum fitness value is taken as the leader, and the remaining network parameter individuals are divided into explorers and followers according to a predetermined proportion; The leader is updated using a neighborhood search algorithm to obtain an updated leader; The explorer is updated using a cooperative search algorithm to obtain an updated explorer; The follower is updated using an adaptive search algorithm based on the leader and the explorer to obtain an updated follower; The target network parameter individual with the maximum fitness value is selected from the updated leader, the updated explorer and the updated follower, and the target network parameter individual is taken as the global optimal value; The updating of the leader, the explorer and the follower is repeated until the maximum training number is reached, and the global optimal value is taken as the final network parameter of the usage prediction model to obtain a trained usage prediction model; During the training process, the leader, the explorer and the follower are re-divided every training cycle, and each training cycle includes multiple training processes; The historical usage records of fracturing proppants corresponding to each target user are collected, and the historical usage records are identified using the trained usage prediction model to obtain the predicted usage of each target user, and the predicted usage of each target user is sent to the work terminal corresponding to the warehousing worker to assist the warehousing worker in restocking; After sending the predicted usage of each target user to the work terminal corresponding to the warehousing worker, real-time warehousing scheduling information transmitted by the work terminal is obtained, and the real-time warehousing scheduling information includes the type and quantity of fracturing proppants; According to the real-time warehousing scheduling information and the current warehousing information, the warehousing robot's picking route is planned, and the warehousing robot is controlled to pick according to the established picking route to complete the fracturing proppant storage scheduling and improve the warehousing efficiency; the current warehousing information includes the type, quantity and shelf number of the warehoused fracturing proppants; The leader is updated using a neighborhood search algorithm to obtain an updated leader, including: The historical optimal value of the leader in the historical training process and the global optimal value in the current updating process are obtained, and a first target vector is obtained by subtracting the global optimal value in the current updating process from the historical optimal value corresponding to the leader; obtaining a norm of the first target vector, and generating a random number according to a normal distribution N (0,1) to obtain the first coefficient α attenuating the norm of the first target vector to obtain the first coefficient After multiplying the global optimal value in the current updating process by a first coefficient, a first updated value is obtained; The first updated value and the global optimal value in the current updating process are added to obtain the updated leader; The explorer is updated using a cooperative search algorithm to obtain an updated explorer, including: The historical optimal value of all explorers in the historical training process is obtained; A target explorer is obtained by randomly selecting one explorer from all explorers; Randomly select an explorer from all explorers except the target explorer, obtain the collaborator corresponding to the target explorer, and determine the historical optimal value corresponding to the collaborator; Get the mean between the historical optimal value of the target explorer and the historical optimal value of the collaborator to obtain the mean vector; On the basis of the current training number, a step factor is obtained, and a random number generated according to a normal distribution N (0,1) α The step factor is attenuated to obtain a second coefficient; After multiplying the mean vector by the second coefficient, a second update value is obtained, and the mean vector and the second update value are added to obtain an updated target explorer; Traverse all explorers and obtain updated explorers; Based on the leader and explorer, an adaptive search algorithm is used to update the followers to obtain updated followers, including: Obtain the leader's reverse learning individuals; Determine the first decision interval (0, 0.33], the second decision interval (0.33, 0.66], and the third decision interval (0.66, 1), and randomly generate a decision factor between (0, 1) for each follower; When the decision factor is in the first decision interval, the follower is guided and updated based on the leader to obtain an updated follower; When the decision factor is in the second decision interval, a random explorer is determined for the follower, and based on the determined explorer, the follower is guided and updated to obtain an updated follower; When the decision factor is in the third decision interval, the follower is guided and updated based on the reverse learning individual to obtain an updated follower; When the decision factor is in the first decision interval, the leader is used as the basis to guide the update of the followers, and the updated followers are: When the decision factor is in the first decision interval, the global optimal value minus the follower's historical optimal value is used to obtain the second target vector; Obtaining a norm of the second target vector, and dividing the norm of the second target vector by the interval length of the network parameter to obtain a third coefficient; wherein the interval length is used to represent the upper limit minus the lower limit of the network parameter; The fourth coefficient is determined based on the follower's fitness during the previous two training processes; Obtaining a first update step factor of the follower according to the third coefficient and the fourth coefficient; random numbers generated according to a normal distribution N (0,1) α a first update step factor for the follower is attenuated to obtain a fifth coefficient Multiply the average of the global optimal value and the follower's historical optimal value by the fifth coefficient to obtain the third update value corresponding to the follower; Add the average of the global optimal value and the follower's historical optimal value to the third updated value to obtain the updated follower; When the decision factor is in the second decision interval, a random explorer is determined for the follower, and the follower is guided and updated based on the determined explorer. The updated follower is: When the decision factor is in the second decision interval, the historical optimal value corresponding to the determined explorer is subtracted from the historical optimal value of the follower to obtain the third target vector; Obtaining a norm of the third target vector and dividing the norm of the third target vector by the interval length of the network parameter to obtain a sixth coefficient; wherein the interval length is used to represent the upper limit minus the lower limit of the network parameter; The seventh coefficient is determined based on the follower's fitness during the previous two training processes; Obtaining a second update step factor of the follower according to the sixth coefficient and the seventh coefficient; random numbers generated according to a normal distribution N (0,1) α a second update step factor for the follower is attenuated to obtain an eighth coefficient; The average value of the historical optimal value of the determined explorer and the historical optimal value of the follower is multiplied by an eighth coefficient to obtain a fourth update value corresponding to the follower; The average value of the historical optimal value of the determined explorer and the historical optimal value of the follower is added to the fourth update value to obtain an updated follower; When the decision factor is located in the third decision interval, the follower is guided to update based on the reverse learning individual to obtain an updated follower: When the decision factor is located in the third decision interval, a fourth target vector is obtained by subtracting the historical optimal value of the follower from the reverse learning individual; The norm of the fourth target vector is obtained, and the norm of the fourth target vector is divided by the interval length of the network parameter to obtain a ninth coefficient; wherein the interval length is used to represent the upper limit minus the lower limit of the network parameter; A tenth coefficient is determined based on the fitness of the follower in the previous two training processes; A third update step factor of the follower is obtained according to the ninth coefficient and the tenth coefficient; random numbers generated according to a normal distribution N (0,1) α a third update step factor for the follower is attenuated to obtain an eleventh coefficient; The average value of the reverse learning individual and the historical optimal value of the follower is multiplied by an eleventh coefficient to obtain a fifth update value corresponding to the follower; The average value of the reverse learning individual and the historical optimal value of the follower is added to the fifth update value to obtain an updated follower.

2. The frac proppant storage scheduling method of claim 1, wherein, The fitness value corresponding to each network parameter individual is obtained, and the network parameter individual with the largest fitness value is taken as the leader, and the remaining network parameter individuals are divided into explorers and followers according to a preset proportion, including: The reciprocal of the error function of the usage prediction model is taken as the fitness function, and the fitness value corresponding to each network parameter individual is obtained based on the fitness function; The network parameter individual with the largest fitness value is taken as the leader, and M network parameter individuals are taken as explorers from the remaining network parameter individuals, and N-1-M network parameter individuals are taken as followers; wherein N represents the total number of network parameter individuals.

3. The method of claim 1, wherein, According to real-time warehouse scheduling information and current warehouse information, the picking route of the warehouse robot is planned, and the warehouse robot is controlled to pick according to the established picking route to complete the storage scheduling of the fracturing proppant, including: According to real-time warehouse scheduling information and current warehouse information, the picking nodes that need to be traversed by the warehouse robot are determined; Based on the picking nodes that need to be traversed by the warehouse robot, a path code is obtained; the first element in the picking path code is the position of the warehouse robot, and the last element in the picking path code is the delivery position; Multiple different picking path codes are repeatedly obtained to obtain multiple picking path codes; The travel time corresponding to each picking path code is obtained, and the travel time is used to represent the time required for the warehouse robot to advance at a constant speed according to the picking path code; The picking path code with the minimum travel time is determined from the multiple picking path codes to obtain a target picking path code; The mutation operation and the crossover operation are performed on the picking path code to obtain an updated picking path code; The target picking path code is re-determined based on the original picking path code and the updated picking path code; If the driving time corresponding to the target pickup path code is not reduced for N times, the target pickup path code is output; otherwise, the half of the codes with the maximum driving time are eliminated from the original pickup path codes and the updated pickup path codes, and the step of performing the mutation operation and the crossover operation is returned to; According to the current storage information, the warehouse robot is controlled to pick up at each pickup node in the order of the target pickup path code, and the fracturing proppant storage scheduling is completed.

4. The frac proppant storage scheduling method of claim 3, wherein, Picking up at each pickup node includes: Scanning the RFID information or the two-dimensional code information of the fracturing proppant storage scheduling to obtain the current storage information of the target warehouse fracturing proppant; Based on the current storage information of the target warehouse fracturing proppant, the warehouse robot is controlled to pick up according to the demand of the real-time storage scheduling information.

Images