Intelligent management method and system for food inventory of large ship kitchen

Abstract

The application relates to the technical field of logistics management, and discloses a food inventory intelligent management method and system for a large ship kitchen, which comprises the following steps: acquiring electric current, refrigerated cabinet temperature, consumption rate, inventory quantity and remaining sailing time, performing time sequence decomposition prediction to obtain a short-term consumption trend; correcting the resistance consumption deviation by using the sailing speed and effective wave height to obtain a corrected consumption trend value; taking the original material list of a supply port, performing differentiated supply-demand matching to obtain a supply information set; screening threshold materials according to the corrected consumption trend value and the supply information set to obtain a category identifier, evaluating the port capacity according to the consumption time length, determining the urgency level, and cross-port optimization distribution to obtain an optimized procurement list; according to the optimized procurement list and the supply information set, the minimum port number is planned to obtain an optimal solution vector, a final procurement scheme is generated according to the port, a ship priority capacity constraint loading simulation is performed, and a logistics management output result is output. The method can realize the precision of food supply decision-making for a large ship kitchen.

Description

Technical Field

[0001] This invention relates to the field of logistics management technology, and in particular to an intelligent food inventory management method and system for large ship kitchens. Background Technology

[0002] Currently, during large ships' ocean voyages, kitchen food inventory needs to be adjusted according to the voyage cycle, crew duties, sea conditions, and resupply conditions. Because food consumption is continuously volatile, and the prices, types, and supply capacity of goods at resupply ports vary, inventory management requires big data management capabilities that include consumption trend forecasting, market information correlation, and coordinated procurement and allocation to achieve dynamic matching between food inventory and resupply nodes.

[0003] In one existing technology, managers establish a kitchen food inventory ledger according to the voyage plan, and manually enter the food categories, remaining quantities, and consumption records periodically. Before arriving at a resupply port, the purchase quantity is estimated based on historical consumption averages and the estimated number of voyage days, and the purchase categories are selected by manually querying the port's supply list and price information. After the purchase is completed, the inventory ledger is updated again, serving as the basis for food rationing and resupply arrangements for the next voyage. This existing technology relies on manual ledgers and historical consumption experience for food inventory estimation, making it difficult to synchronously correlate consumption fluctuations with the prices, categories, and supply capacity of the resupply port.

[0004] Therefore, existing technologies cannot achieve precise decision-making for food replenishment in the kitchens of large ships. Summary of the Invention

[0005] This invention provides a method and system for intelligent food inventory management in the kitchens of large ships, so as to achieve precise food replenishment decisions in the kitchens of large ships.

[0006] In a first aspect, to solve the above-mentioned technical problems, the present invention provides a method for intelligent management of food inventory in the kitchen of large ships, comprising: By acquiring electrical current, refrigerator temperature, material consumption rate, material inventory balance and remaining sailing time, time-series trend decomposition and prediction are performed to obtain short-term consumption trend values. The ship's speed and significant wave height are obtained, and the short-term consumption trend value is corrected for resistance consumption deviation trend to obtain the corrected consumption trend value. Obtain the original material list of the supply port, conduct differentiated supply and demand matching analysis, and obtain a set of supply information; Based on the corrected consumption trend value and the supply information set, trigger threshold materials are screened to obtain category identifiers. Based on the category identifiers, port capacity assessment based on consumption time is performed to obtain supply urgency level. Based on the supply urgency level, cross-port procurement volume is optimized and allocated to obtain an optimized procurement list. The optimal solution vector is obtained by constructing a minimum number of ports based on the optimized procurement list and the supply information set, and the final procurement plan is generated according to the port procurement instructions based on the optimal solution vector. Based on the final procurement plan, a ship priority capacity constraint loading simulation was performed to obtain the logistics management output results.

[0007] Secondly, the present invention provides an intelligent food inventory management system for a large ship kitchen, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the method described above.

[0008] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described above.

[0009] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention obtains short-term consumption trend values ​​by acquiring electrical current, refrigerated cabinet temperature, material consumption rate, material inventory balance, and remaining sailing time, and then performs time-series trend decomposition prediction. The short-term consumption trend values ​​are then corrected for resistance consumption deviation using ship speed and significant wave height to obtain corrected consumption trend values. Since time-series decomposition can separate the periodic and trend components in the consumption data, and the resistance correction model dynamically adjusts the consumption rate based on real-time speed and wave height, this invention solves the problem in existing technologies where consumption prediction relies solely on historical averages and cannot respond to changes in sea state and equipment status fluctuations, thereby improving the dynamic adaptability and accuracy of consumption trend prediction.

[0010] 2. This invention obtains a supply information set by acquiring the original material list of the supply port and performing differentiated supply and demand matching analysis. Then, based on the corrected consumption trend value and the supply information set, it filters materials according to trigger thresholds to obtain category identifiers. Based on the category identifiers, it performs port capacity assessment based on consumption time to obtain the supply urgency level. Subsequently, it optimizes the allocation of cross-port procurement volume based on the supply urgency level to obtain an optimized procurement list. Since differentiated supply and demand matching identifies materials with different names and substitute materials, and the supply urgency level quantifies the relationship between the remaining available days and the port transfer time, cross-port allocation prioritizes matching scarce materials to the port with the fastest response. This solves the problem in the prior art that manual querying of port lists cannot dynamically associate consumption urgency with differences in supply capacity, and achieves accurate supply and demand matching and optimized allocation of procurement volume.

[0011] 3. This invention constructs and solves the optimal solution vector by optimizing the procurement list and supply information set to minimize the number of ports. Based on the optimal solution vector, it generates the final procurement plan by port. Then, based on the final procurement plan, it performs a ship priority capacity constraint loading simulation to obtain the logistics management output. Since the minimum number of ports planning aims to minimize Boolean variables and constructs a linear programming model by combining supply capacity and material gaps, it can select the fewest supply ports to cover all procurement needs from a global perspective. The priority capacity constraint loading simulation performs three-dimensional spatial loading according to material priority, ship remaining load capacity and volume and generates action code instructions. This solves the problems of the procurement plan being disconnected from the actual loading capacity of the ship and the difficulty of coordinating multi-port supply in the prior art. It reduces the number of supply ports and logistics complexity and ensures the feasibility of the procurement plan under ship capacity constraints. Attached Figure Description

[0012] Figure 1 This is a schematic diagram of the intelligent food inventory management method for large ship kitchens provided in the first embodiment of the present invention. Detailed Implementation

[0013] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0014] Reference Figure 1 The first embodiment of the present invention provides a method for intelligent management of food inventory in the kitchen of a large ship, comprising the following steps: S11: Obtain electrical current, refrigerator temperature, material consumption rate, material inventory balance and remaining sailing time, perform time-series trend decomposition and prediction, and obtain short-term consumption trend value. S12, obtain the ship speed and significant wave height, and correct the resistance consumption deviation trend value of the short-term consumption trend value to obtain the corrected consumption trend value; S13, Obtain the original material list of the supply port, perform differentiated supply and demand matching analysis, and obtain a set of supply information; S14. Based on the corrected consumption trend value and the supply information set, trigger threshold material screening is performed to obtain category identifiers. Based on the category identifiers, port capacity assessment based on consumption time is performed to obtain supply urgency level. Based on the supply urgency level, cross-port procurement volume is optimized and allocated to obtain an optimized procurement list. S15, Based on the optimized procurement list and the supply information set, the minimum number of ports is planned and constructed to obtain the optimal solution vector, and the final procurement plan is generated according to the port procurement instructions based on the optimal solution vector. S16. Based on the final procurement plan, perform a ship priority capacity constraint loading simulation to obtain the logistics management output results.

[0015] In step S11, electrical current, refrigerator temperature, material consumption rate, material inventory balance, and remaining sailing time are acquired, and time-series trend decomposition prediction is performed to obtain short-term consumption trend values, including: The electrical current, the refrigerator temperature, the material consumption rate, the remaining material inventory, and the remaining sailing time are time-aligned to generate a multi-dimensional synchronization sequence; The multidimensional synchronization sequence is split into periodic components, trend components and residual components using the seasonal trend decomposition method, and the trend components are extracted as consumption feature components. Based on the consumed feature components, a sliding window is used to extract feature data within a preset window to construct a time window feature matrix; Based on the time window feature matrix, input the data into the Long Short-Term Memory network for state memory and normalized prediction, and output the intermediate prediction value. Based on the predicted intermediate value, it is denormalized to restore the predicted consumption value, and then superimposed with the preset baseline consumption value to obtain the short-term consumption trend value.

[0016] In step S11, the process of obtaining electrical current, refrigerated container temperature, material consumption rate, material inventory balance, and remaining sailing time, and performing time-series trend decomposition prediction to obtain short-term consumption trend values ​​is as follows: The electrical current value (in amperes) on the power supply line of the refrigerated equipment is read from the ship's galley power monitoring system; the refrigerated container temperature value (in degrees Celsius) is read from the built-in temperature sensor; and the material consumption rate (in kilograms per day) for each type of material is obtained from the ship's logistics management system. This rate is calculated by dividing the total consumption over the past 7 consecutive days by 7. For example, if the total consumption over the past 7 days was 140 kilograms, the consumption rate is 20 kilograms per day. The current material inventory balance (in kilograms) is read from the ship's inventory ledger. For example, if the current inventory balance is 500 kilograms. The remaining sailing time (in days) is calculated from the ship's sailing plan based on the distance between the current ship position and the next resupply port and the planned speed. For example, if the distance is 1200 nautical miles and the planned speed is 15 knots, the remaining sailing time is 3.3 days. The above five values ​​were collected every hour at a uniform time interval, for a total of 72 sets of data collected over the past 72 hours. Each set of data contained the five values ​​at the same time. These 72 sets of data were arranged in chronological order to generate a 72-row, 5-column multidimensional synchronization sequence. Each column corresponds to the electrical current, refrigerator temperature, material consumption rate, material inventory balance, and remaining sailing time.

[0017] Among them, the electrical current and the refrigerator temperature are used to detect abnormal operating conditions of the refrigeration equipment. When the electrical current deviates from the normal range by more than 20% (the preset normal range is 20-30 amps), or the refrigerator temperature is higher than 5°C, the refrigeration equipment is judged to be faulty. At this time, the material consumption rate is multiplied by a compensation coefficient of 1.2 to simulate the additional consumption caused by accelerated food spoilage. The normal range of 20-30 amps is obtained based on the nameplate parameters of the ship's refrigeration equipment and historical normal operation data. The temperature threshold of 5°C is determined based on the refrigeration temperature requirements in the food safety standards. The compensation coefficient of 1.2 is obtained based on the average increase in the food consumption rate during historical fault periods.

[0018] A seasonal trend decomposition method based on locally weighted regression was used to process this multidimensional synchronous sequence. For the 72 values ​​in the material consumption rate column, the periodic component was first extracted. The period length was set to 24 hours, and the arithmetic mean of the consumption rate 12 times before and after each time point was calculated to obtain the periodic component of the 24-hour repeating fluctuation. For example, the periodic component at the 30th hour is equal to the average consumption rate of the other 24 times between the 18th and 42nd hours, excluding that time. The periodic component was subtracted from the original consumption rate sequence to obtain the sequence after removing the period. Locally weighted regression was used on this sequence to extract the trend component. The locally weighted regression takes a window with a width of 48 hours around each data point, that is, the window contains data from 24 times before and after the current point. Each data point in the window is assigned a weight, with the weight increasing the closer to the current point. A cubic kernel function was used to calculate the weights. Then, a weighted linear fit was performed within the window, and the value on the smooth curve obtained by the fit is the trend component. The residual component was obtained by subtracting the periodic component and the trend component from the original consumption rate sequence. The trend component is extracted as the consumption feature component. This component contains the slow change pattern caused by changes in the number of crew members or changes in the latitude of the route, and eliminates daily periodic fluctuations and random noise.

[0019] A time window feature matrix is ​​constructed based on the consumption feature components. First, the length of the sliding window needs to be determined, which is done by statistically analyzing historical data. One hundred consumption feature subsequences of 3 days each are randomly selected from the past year's flight records. Each subsequence contains consumption feature values ​​at 72 moments. The Pearson correlation coefficient between each subsequence and the actual consumption values ​​at 24 moments in the following day is calculated. The 90th percentile of the subsequence length corresponding to the first drop below 0 is taken as the window width, which is statistically determined to be 48 moments (e.g., the 90th value after sorting is 48). Then, the sliding window is moved along the time axis of the consumption feature components with a step size of 1 hour. Each time the window slides, it extracts the consumption feature values ​​at the current moment and the previous 47 moments, resulting in 48 values ​​arranged in a column, forming a 48-row, 1-column time window feature matrix.

[0020] The temporal window feature matrix is ​​input into the Long Short-Term Memory (LSTM) network for prediction. Before training, a training sample set needs to be constructed, consisting of consumption feature data recorded from three complete voyages. Each voyage lasts 30 days, totaling 2160 time points, for a total of 6480 time points across the three voyages. Input sequences and output labels are extracted from this data. Each input sequence represents the consumption feature values ​​for 48 consecutive time points, and the corresponding output label represents the actual consumption value at the next immediate time point. Max-min normalization is performed on each consumption feature value in the input sequence. The global maximum and global minimum values ​​of all consumption feature values ​​in the training sample set are calculated; for example, the global maximum is 30 kg / day and the global minimum is 10 kg / day. Each original consumption feature value is subtracted from the global minimum and then divided by the difference between the global maximum and global minimum values ​​to obtain a normalized value ranging from 0 to 1. The network structure employs a two-layer stacked LSM network. The first layer has 50 memory units, and the second layer has 30 memory units. Each memory unit contains a forget gate, an input gate, and an output gate. The forget gate processes the previous hidden state and the current normalized input using a logistic function, outputting a value between 0 and 1 that determines the proportion of the previous cell state retained. The input gate similarly uses a logistic function to determine the update weights for new information, then creates candidate cell states using a tangent function. The update weights are multiplied by the candidate cell states and added to the previous cell state filtered by the forget gate to obtain the current cell state. The output gate processes the previous hidden state and the current input using a logistic function, then multiplies by the value of the current cell state after the tangent function to obtain the current hidden state. The output of the second layer is fed into a fully connected layer containing one neuron, outputting a normalized prediction intermediate value. Network training uses mean squared error as the loss function, employs an adaptive moment estimation optimizer, a learning rate of 0.001, a batch size of 32, and 100 training iterations. After training, the time window feature matrix constructed in step S11 is input into the network. The network processes 48 input values ​​sequentially at each time step, finally outputting a normalized prediction intermediate value, a real number between 0 and 1, for example, 0.65.

[0021] The normalized prediction median is used for inverse normalization. The inverse normalization calculation formula is: the predicted consumption value equals the normalized prediction median multiplied by the difference between the global maximum and global minimum values ​​in the training sample set, plus the global minimum value. For example, if the normalized prediction median is 0.65, the global maximum of 30 kg / day minus the global minimum of 10 kg / day equals 20, then 0.65 multiplied by 20 equals 13, plus 10 equals 23 kg / day. This predicted consumption value reflects the future short-term consumption amount extrapolated from the trend component. The short-term consumption trend value is obtained by superimposing this predicted consumption value with a preset baseline consumption value. The baseline consumption value is determined as follows: the actual daily consumption values ​​are extracted from the past 5 voyages, with each voyage lasting 30 days, for a total of 150 actual consumption values. These values ​​are sorted from smallest to largest, and the value corresponding to the 10th percentile is taken as the baseline consumption value. For example, if the 15th value after sorting is 18 kg / day, then the baseline consumption value is 18 kg / day. The overlay operation involves adding the predicted consumption value to the baseline consumption value; for example, 23 plus 18 equals 41 kg per day, ultimately yielding the short-term consumption trend value. This step incorporates electrical current and refrigerator temperature to reflect the indirect impact of refrigeration equipment operating conditions on food consumption. It utilizes seasonal trend decomposition to separate cyclical fluctuations from long-term trends, and then employs a long short-term memory network to capture nonlinear time-series characteristics. This allows the short-term consumption trend value to dynamically adapt to changes in sea conditions and navigation status, providing a more accurate numerical basis than historical averages for subsequent replenishment trigger judgments.

[0022] In step S12, the ship's speed and significant wave height are obtained, and the short-term consumption trend value is corrected for resistance consumption deviation trend to obtain a corrected consumption trend value, including: The ship's resistance variables are calculated by inputting the ship's speed and effective wave height into the ship resistance increment model. Divide the resistance variable by the preset standard resistance to obtain the consumption rate correction coefficient; Multiply the consumption rate correction factor by the material consumption rate to obtain the consumption deviation value; The consumption deviation value is superimposed on the short-term consumption trend value to obtain the corrected consumption trend value.

[0023] Specifically, ship speed is measured in knots, obtained in real-time from the ship's logbook or GPS receiver. Significant wave height is measured in meters, obtained from the ship's wave radar or wave monitoring sensors. Significant wave height is defined as the average of the top third of the largest wave heights observed over the past 30 minutes, sorted from largest to smallest. The ship resistance increment model is a pre-built mathematical model based on ship hydrodynamics principles. Thirty sets of data recorded during the ship's voyages over the past 12 months were collected. Each set includes ship speed, significant wave height, and the corresponding actual total ship resistance increment. The actual total ship resistance increment is calculated by inversely measuring the relationship between the ship's main engine power and speed. Specifically, it is calculated by subtracting the current actual main engine power from the main engine power under standard calm water conditions at the same speed, dividing by the main engine power under standard conditions, and then multiplying by the standard resistance to obtain the resistance increment.

[0024] The ship speed range in the above 30 data sets is divided into intervals of 0.5 knots, and the significant wave height range is divided into intervals of 0.2 meters. The arithmetic mean of the resistance increments within each interval is taken as the corresponding model output value. The final constructed ship resistance increment model is represented by a two-dimensional lookup table. The input is the ship speed and significant wave height, and the output is the ship resistance variable in Newtons. When using this model, the currently read ship speed value is rounded to the nearest 0.5 knot, and the current significant wave height value is rounded to the nearest 0.2 meter. The corresponding resistance increment value is located from the lookup table as the ship resistance variable output. For example, a ship speed of 12.3 knots rounded to 12.5 knots and a significant wave height of 1.7 meters rounded to 1.8 meters result in a ship resistance variable of 15,000 Newtons.

[0025] The consumption rate correction factor is obtained by dividing the ship's resistance variable by the preset standard resistance. The preset standard resistance is determined as follows: the total resistance value of the ship under design draft, calm water conditions, and economic speed is obtained from the ship's design manual, and this resistance value is used as the standard resistance. If the design manual is unavailable, historical data is used to select all navigation periods from the past 12 months where the effective wave height is less than 0.5 meters and the wind speed is less than level 3. During these periods, main engine power data at economic speed is collected, and the average resistance value is calculated using the conversion formula between main engine power and resistance. This average resistance value is used as the standard resistance, for example, 30,000 Newtons. The consumption rate correction factor is calculated by dividing the ship's resistance variable by the standard resistance, resulting in a dimensionless real number, for example, 15,000 divided by 30,000 equals 0.5.

[0026] The consumption rate correction factor is multiplied by the material consumption rate to obtain the consumption deviation value. The material consumption rate is derived from the material consumption rate value obtained in step S11, and the unit is kilograms per day. The consumption deviation value is calculated as follows: consumption rate correction factor multiplied by material consumption rate. For example, if the consumption rate correction factor is 0.5 and the material consumption rate is 20 kilograms per day, then the consumption deviation value is 10 kilograms per day. This consumption deviation value reflects the additional increase in actual consumption relative to standard conditions due to the current sea state and speed.

[0027] The consumption deviation value is superimposed on the short-term consumption trend value to obtain the corrected consumption trend value. The short-term consumption trend value is derived from the output of step S11, for example, a short-term consumption trend value of 41 kg per day. The superposition operation is: the short-term consumption trend value is added to the consumption deviation value, for example, 41 plus 10 equals 51 kg per day, resulting in the corrected consumption trend value. This step introduces two dynamic variables, ship speed and significant wave height, and uses a ship resistance increment model to quantitatively calculate the impact of sea state and speed on ship navigation resistance, which is then converted into a correction amount for the food and material consumption rate. This ensures that the corrected consumption trend value can reflect the increase in food consumption caused by increased energy demand from crew members under adverse sea conditions or high-speed navigation conditions, thus avoiding the risk of material shortages due to sudden changes in sea state when relying solely on historical average consumption data for replenishment decisions. The corrected consumption trend value serves as the basis for determining the replenishment trigger threshold and calculating the remaining available days in subsequent steps, and its accuracy directly determines the reliability of the replenishment decision.

[0028] In step S13, the original material list of the supply port is obtained and a differential supply and demand matching analysis is performed to obtain a supply information set, including: Obtain the original material list of the supply port, identify the name differences and substitutes in the original material list, and generate a differentiated material catalog; Based on the differentiated material catalog, the upper limit of the supply capacity of the supply port and the time required for cargo transfer are obtained, thus obtaining the set of port supply capacity. Multiplying the resource consumption rate by the remaining sailing time yields the estimated total consumption. Subtracting the current inventory balance from the estimated total consumption gives the resource shortage figure. Based on the port supply capacity set and the material shortage value, the available material quantity for supplying the port is obtained, and a supply information set containing category differences and supply capacity information is generated.

[0029] Specifically, obtain the name and estimated arrival time of the next replenishment port from the ship automatic identification system or voyage plan, send a request to the material supplier system of the port via a satellite communication link, and receive the returned original material list file, which contains information such as material name, specification, packaging form, and inventory quantity. There may be differences between the material names in the original material list and those used inside the ship's kitchen. For example, "potato" is used inside the ship while "potato" is used in the port list, or the port provides "frozen chicken drumsticks" while the ship can accept "frozen whole chicken legs" as a substitute. To identify the materials with name differences and the substitutable materials, a material name mapping database is pre-constructed. The construction process of this database is as follows: Collect the original material lists of 20 different ports where the ship has docked in the past 3 years, with a total of about 2,000 material records, and at the same time collect a list of standard material names used inside the ship's kitchen, with a total of 150 categories. For each port material record, calculate the string edit distance between its name and each standard material name. The edit distance is defined as the minimum number of single-character edit operations required to convert one string to another. For example, to convert "potato" to "potato", it is necessary to replace "horse" with "soil" and "bell" with "bean", for a total of 2 operations. Take the standard material name with the minimum edit distance as the matching candidate, and establish a mapping relationship if the edit distance is less than or equal to 3. For materials that cannot be matched by the edit distance, manually mark their substitutable relationship with the standard materials. For example, "frozen chicken drumsticks" can substitute for "frozen whole chicken legs" but cannot substitute for "chicken breast". Store all the automatically matched and manually marked mapping relationships as the material name mapping database. This database contains two columns, one is the original port material name, and the other is the corresponding ship standard material name. When using this database, use each material name in the original material list as a query key and perform an exact match in the database. If the match is successful, output the corresponding ship standard material name; if the match fails, mark it as an unrecognized material and do not include it in the subsequent processing for the time. After the above matching, gather all the successfully matched material entries to generate a differential material catalog, which contains the original port material name, the corresponding ship standard material name, the inventory quantity in the port list, and the material specification unit.

[0030] Based on the differentiated materials catalog, obtain the upper limit of supply capacity and the time required for cargo relocation for each supply port. The upper limit of supply capacity is obtained as follows: for each material in the catalog, query the port supplier system for the current available quantity of that material, and use this quantity as the upper limit of supply capacity, in kilograms or pieces. If the supplier system does not directly provide the available quantity, subtract the quantity locked to other vessels from the inventory quantity in the original materials list to obtain the remaining available quantity as the upper limit of supply capacity. The time required for cargo relocation is obtained as follows: for each material, query the port supplier system for the shortest preparation time from order confirmation to the arrival of the goods at the terminal loading / unloading point, in hours. If the supplier system does not provide this data, use the default value based on the material category: 6 hours for fresh materials, 12 hours for frozen materials, and 24 hours for dry goods. By aggregating the upper limit of supply capacity and the time required for cargo transfer for each type of material in the differentiated material catalog, a port supply capacity set is obtained. This set is indexed by the standard material name of the ship and includes two attributes: upper limit of supply capacity and time required for cargo transfer.

[0031] The estimated total consumption is obtained by multiplying the resource consumption rate by the remaining voyage time. The resource consumption rate is derived from the resource consumption rate value obtained in step S11, and the unit is kilograms per day. The remaining voyage time is derived from the remaining voyage time value obtained in step S11, and the unit is days. The estimated total consumption is calculated by multiplying the resource consumption rate by the remaining voyage time. For example, if the resource consumption rate is 20 kilograms per day and the remaining voyage time is 3.3 days, the estimated total consumption is 66 kilograms. Then, the estimated total consumption is subtracted from the current inventory balance to obtain the resource shortage value. The current inventory balance is derived from the resource inventory balance value obtained in step S11, and the unit is kilograms. For example, if the current inventory balance is 500 kilograms. If the result is positive, it indicates that there is a shortage. If the result is less than or equal to 0, the resource is marked as not needing replenishment and will not proceed to the subsequent threshold screening step. For example, 66 minus 500 equals -434, which means there is no shortage. For example, assuming a resource consumption rate of 20 kg per day and a remaining sailing time of 30 days, the estimated total consumption is 600 kg, the current inventory balance is 200 kg, and the resource shortage is 400 kg. The unit of the resource shortage value is kilograms.

[0032] Based on the port supply capacity set and the material shortage value, the available quantity of materials to be supplied to the port is obtained. Specifically, for each material with a shortage, the upper limit of the supply capacity for that material in the port supply capacity set is compared with the material shortage value, and the smaller value is taken as the available quantity of materials that the port can actually procure. For example, if the material shortage value is 400 kg and the port supply capacity upper limit is 300 kg, then the available quantity is 300 kg; if the port supply capacity upper limit is 500 kg, then the available quantity is 400 kg. The available quantity of each material, the time required for relocation, and the correspondence between the original name and standard name in the differentiated material catalog are summarized to generate a supply information set. This supply information set uses the ship's standard material name as the primary key and includes four fields: port original material name, upper limit of supply capacity, available quantity of material, and time required for relocation. This step solved the naming inconsistency between ports and ships by establishing a material name mapping database, incorporated port supply-side constraints into management by quantifying the upper limit of supply capacity and the time required for cargo transfer, clarified the actual demand quantity by calculating the material gap value, and finally generated a set of replenishment information to provide complete supply and demand matching basic data for subsequent steps such as trigger threshold screening, urgency assessment and procurement quantity optimization, ensuring that replenishment decisions can be differentiated according to the actual supply conditions of each port.

[0033] In step S14, trigger threshold material screening is performed based on the corrected consumption trend value and the supply information set to obtain category identifiers. Port capacity assessment based on consumption duration is then performed using these category identifiers to obtain a supply urgency level. Finally, cross-port procurement volume optimization allocation is performed based on the supply urgency level to obtain an optimized procurement list, including: Extract the category identifier of the material corresponding to the corrected consumption trend value being greater than the preset supply trigger threshold; Based on the category identifier, retrieve the port availability of the corresponding material at the supply port and the time required for the transfer from the supply information set; Divide the remaining inventory of materials by the corrected consumption trend value to obtain the remaining available days, and determine the supply urgency level according to the remaining available days in ascending order; Based on the supply urgency level and the available quantity at the port, the materials with the highest urgency are prioritized and allocated to the supply port with the shortest relocation time. The procurement volume of each supply port is then allocated according to the material priority to generate an optimized procurement list.

[0034] Specifically, firstly, the corrected consumption trend value for each material is obtained from step S12, in kilograms per day. For example, the corrected consumption trend value for a certain material is 51 kilograms per day. The preset replenishment trigger threshold is a pre-set value, determined by collecting 15 records of replenishment shortages from the ship's voyage records over the past 12 months. The 90th percentile of the actual consumption rate at the time of each shortage is taken as the replenishment trigger threshold, in kilograms per day. For example, the calculated threshold is 45 kilograms per day. The corrected consumption trend value for each material is compared with the replenishment trigger threshold, and the category identifiers corresponding to materials with corrected consumption trend values ​​greater than the replenishment trigger threshold are extracted. The category identifier is a unique code for each material; for example, rice is coded as R001, and flour as F002. After filtering, a list of category identifiers for materials to be replenished is obtained.

[0035] Based on the selected category identifiers, the available quantity of the corresponding material at each supply port and the required transfer time are retrieved from the supply information set generated in step S13. The supply information set contains supply information for each material at each of the upcoming supply ports. Assume there are three supply ports: Port A, Port B, and Port C. For rice with category identifier R001, the available quantity at Port A is 300 kg, and the transfer time is 6 hours; at Port B, it is 500 kg, and the transfer time is 12 hours; and at Port C, it is 200 kg, and the transfer time is 24 hours. The retrieved data is organized by material.

[0036] The remaining usable days are calculated by dividing the remaining inventory by the adjusted consumption trend value. The remaining inventory is obtained from step S11. For example, if the current remaining rice inventory is 200 kg and the adjusted consumption trend value is 51 kg per day, then the remaining usable days are approximately 3.9 days (200 divided by 51). The remaining usable days are calculated for each type of supply to be replenished. The urgency level is determined based on the remaining usable days, from smallest to largest. Specifically, the remaining usable days are divided into three levels: Level 1 urgency (less than or equal to 2 days), Level 2 urgency (greater than 2 days but less than or equal to 5 days), and Level 3 urgency (greater than 5 days). For example, rice with 3.9 remaining days is Level 2 urgency, while eggs with 1.5 remaining days are Level 1 urgency. A smaller urgency level indicates a higher level of urgency.

[0037] The cross-port procurement volume is optimized and allocated based on the supply urgency level and port availability. First, all supplies awaiting replenishment are sorted from highest to lowest urgency level, with Level 1 urgency supplies at the top, followed by Level 2, and then Level 3. Within the same urgency level, supplies are further sorted by remaining days of availability from smallest to largest. Then, each type of supply is processed sequentially according to this priority order. For the currently processed supply, the port with the shortest relocation time is selected as the primary procurement port. If the available quantity at this port is greater than or equal to the supply shortage value, the entire shortage quantity is procured at that port, marking the allocation of the supply complete. If the available quantity at this port is less than the supply shortage value, the entire available quantity is procured at that port, and then the port with the second shortest relocation time is selected from the remaining ports to continue procuring the remaining shortage until the shortage is completely satisfied or all port availability is exhausted. For example, if a material shortage is 400 kg, and port A can obtain 300 kg with a transfer time of 6 hours, while port B can obtain 200 kg with a transfer time of 12 hours, then 300 kg will be purchased from port A first, and the remaining 100 kg will be purchased from port B. The purchase volume of each supply port is allocated according to the material priority, ultimately generating an optimized purchase list. The optimized purchase list is stored in tabular form, including fields such as material category identifier, port name, purchase quantity at that port, and purchase sequence number. This step automatically filters out materials with supply risks by setting a supply trigger threshold based on historical shortage statistics, then quantifies the urgency using the remaining available days, and finally allocates the purchase volume to the most suitable port according to the principles of urgency priority and transfer time priority. This avoids the subjectivity of manual judgment, ensures that scarce materials can be replenished from the fastest responding port, and provides an initial purchase volume allocation scheme for subsequent minimum port number planning.

[0038] In step S15, the minimum port number planning is performed based on the optimized procurement list and the supply information set to obtain the optimal solution vector. Then, the final procurement plan is generated according to the port procurement instructions based on the optimal solution vector, including: Based on the optimized procurement list and the supply information set, a Boolean variable is introduced for the supply port to indicate whether to perform procurement at the supply port. With the goal of minimizing the sum of all the Boolean variables, the procurement quantity of materials at the supply port is used as a continuous decision variable to construct a linear programming model. The linear programming model is input into a preset mixed-integer linear programming solver for iterative solution to obtain the optimal solution vector for replenishing port procurement volume; Based on the optimal solution vector, independent procurement instructions are generated according to the allocation of supply ports, and a final procurement plan is determined, including the procurement port, the category of procurement materials, the weight of procurement materials, the total volume of procurement materials, and the total weight of procurement materials.

[0039] The constraints of the linear programming model are as follows: if the procurement quantity at the supply port is greater than zero, it must not be less than the preset minimum order quantity and must not exceed the upper limit of the supply capacity; the sum of the procurement quantities at all supply ports equals the material shortage value.

[0040] Specifically, the optimized procurement list is derived from the output of step S14. This list includes the category identifier of each material, the planned procurement quantity at each supply port, and the procurement sequence number. The supply information set is derived from the output of step S13. This set includes the upper limit of the supply capacity of each supply port for each material and the time required for relocation. Assume there are currently 3 candidate supply ports, denoted as Port X, Port Y, and Port Z, and 2 types of materials to be supplied, denoted as Material A and Material B. A Boolean variable is introduced for each supply port. This variable can only take the value 0 or 1. A value of 1 indicates that procurement will be carried out at this port, and a value of 0 indicates that procurement will not be carried out at this port. At the same time, for each port-material combination, the procurement quantity must not exceed the upper limit of the supply capacity multiplied by the Boolean variable of that port. That is, when the Boolean variable is 0, the procurement quantity must be 0. If the procurement quantity is greater than 0, it must also not be less than the preset minimum order quantity, such as 50 kg. This constraint can be achieved by introducing an auxiliary Boolean variable or using a semi-continuous variable. Minimizing the sum of all Boolean variables is the objective function of the linear programming model, which aims to use as few supply ports as possible to complete all procurement tasks.

[0041] The actual purchase quantity of each supply port for each type of material is treated as a continuous decision variable, which is a non-negative real number in kilograms. For material A, the decision variables include the purchase quantity of material A by port X, port Y, and port Z; the same applies to material B. The linear programming model needs to satisfy the following constraints. The first constraint is that for each combination of supply port and each type of material, if the purchase quantity of that port is greater than 0, then the purchase quantity must not be less than the preset minimum order quantity and must not exceed the upper limit of the port's supply capacity.

[0042] The preset minimum order quantity is determined by extracting all order data from the ship's actual procurement records at all supply ports over the past two years, totaling 200 records. The procurement weight of each material in each order is calculated, and all material procurement weights are sorted from smallest to largest. The value corresponding to the 10th percentile is taken as the minimum order quantity. For example, if the 20th percentile value is 50 kg, then the minimum order quantity is 50 kg. For port X's procurement quantity of material A, if it is greater than 0, it must be greater than or equal to 50 kg and less than or equal to port X's upper limit of supply capacity for material A (e.g., 300 kg). The second constraint is that for each material, the sum of the procurement quantities at all supply ports must equal the material shortage value for that material. The material shortage value comes from step S13. For example, if the shortage of material A is 400 kg, then the sum of the procurement quantities of material A from ports X, Y, and Z must equal 400 kg. The objective function and the above constraints are organized in matrix form to construct a complete linear programming model.

[0043] The linear programming model is input into a pre-defined mixed-integer linear programming solver for iterative solution. The mixed-integer linear programming solver is a pre-written computer program module that uses a branch-and-bound method combined with the simplex algorithm for solution. The solver's input parameters include the coefficient vector of the objective function, the coefficient matrix of the constraints, the constant vector on the right-hand side of the constraints, and the boundary vectors of the decision variables. In the objective function coefficient vector, the coefficient of each Boolean variable is 1, and the coefficient of each continuous decision variable is 0, because the minimization objective only concerns the number of ports activated and not the purchase quantity. The coefficient matrix of the constraints consists of the coefficients corresponding to each constraint. The solver executes according to the iterative steps of the simplex algorithm. First, it finds an initial basic feasible solution, then it sequentially replaces basic and non-basic variables along the descent direction of the objective function. Each iteration calculates the test number and updates the step size until all test numbers are non-negative, at which point the iteration stops, yielding the optimal solution. The solver outputs an optimal solution vector containing the optimal values ​​of all Boolean variables and continuous decision variables. For example, in the Boolean variables, port X is 1, port Y is 0, and port Z is 1, indicating that procurement is performed at ports X and Z. In the continuous decision variables, port X purchases 250 kg of material A, port Z purchases 150 kg of material A, port X purchases 100 kg of material B, and port Z purchases 50 kg of material B.

[0044] Based on the optimal solution vector, independent procurement instructions are generated according to the allocation quantities at each supply port. For each selected port to execute the procurement, the types of materials with a value greater than 0 and their procurement quantities are extracted from all continuous decision variables corresponding to that port. This is then summarized to obtain a list of procurement material categories for that port, the procurement weight of each material, the total weight of all procurement materials for that port, and the volume of each material. The material volume is obtained through material density conversion; for example, the unit weight volume parameter of each material is pre-stored in the material attribute database, and the total procurement volume is obtained by multiplying the procurement weight by the unit weight volume. The final procurement plan is a summary document containing a list of procurement port names, a list of procurement material categories corresponding to each procurement port, the procurement weight of each material, the total weight of procurement materials for each port, and the total volume of procurement materials for each port. For example, port X's procurement plan involves purchasing 250 kg of material A and 100 kg of material B. Material A has a unit weight volume of 0.0012 cubic meters per kg, with a total volume of 0.3 cubic meters. Material B has a unit weight volume of 0.0015 cubic meters per kg, with a total volume of 0.15 cubic meters. The total volume of materials procured by port X is 0.45 cubic meters. This step uses minimum port number planning to compress the procurement demand to the fewest supply ports, reducing the complexity of multi-port coordination. Simultaneously, linear programming ensures the globally optimal allocation of procurement quantities under supply capacity and minimum order quantity constraints. The generated final procurement plan clearly defines the detailed procurement tasks for each port, providing accurate material weight and volume parameters for subsequent loading simulations.

[0045] In step S16, a ship priority capacity constraint loading simulation is performed based on the final procurement plan to obtain logistics management output results, including: Based on the final procurement plan, cargo space priorities are assigned to the procured materials. When the total weight of the purchased materials exceeds the pre-obtained remaining deadweight of the ship, materials with lower priority are eliminated sequentially until the total weight of the purchased materials does not exceed the remaining deadweight of the ship. When the total volume of the purchased materials exceeds the pre-obtained remaining volume of the ship, materials with lower priority are eliminated sequentially until the total volume of the purchased materials does not exceed the remaining volume of the ship, thus obtaining the material loading list. Input the material category, weight and volume parameters in the material loading list into the three-dimensional space packing simulator, and generate a cargo loading diagram according to the rule of placing dense materials at the bottom and fragile materials at the top. Extract the storage location coordinates and unloading sequence number of each material from the cargo loading diagram to obtain the supply task queue, and generate action coding instructions containing action sequence number, material identifier, source hold coordinates and target exit position for the unloading action based on the supply task queue. After summarizing all the action coding instructions, simulate their execution and calculate the remaining material inventory and the ship's center of gravity offset after execution to obtain the logistics management output results.

[0046] Specifically, in step S16, the process of simulating ship priority capacity constraint loading based on the final procurement plan to obtain the logistics management output is as follows. The final procurement plan is derived from the output of step S15. This plan includes a list of material categories for each procurement port, the procurement weight of each material, the total volume of the procured materials, and the total weight of the procured materials. First, according to the final procurement plan, hold priorities are set for the procured materials. The rules for setting hold priorities are based on material attributes and historical statistics. The average daily consumption of each material is calculated from the ship's galley material consumption records over the past 12 months. The materials are sorted from highest to lowest daily consumption, with the top 30% set as high priority, the middle 40% as medium priority, and the bottom 30% as low priority. In addition, perishable materials such as fresh vegetables and meat are set as high priority regardless of their consumption level to ensure they are stored in refrigerated warehouses as soon as possible. For materials with long shelf life such as rice and flour, priority is determined based on consumption. Hold priorities are divided into three levels: value 1 represents high priority, value 2 represents medium priority, and value 3 represents low priority.

[0047] Next, we will perform constraint checks on weight and volume. We will obtain the ship's current remaining deadweight in kilograms from the ship's stability report. The remaining deadweight is calculated by subtracting the total weight of currently loaded cargo from the ship's maximum permissible deadweight. We will also obtain the ship's current remaining volume in cubic meters. The remaining volume is calculated by subtracting the currently occupied volume from the total design volume of the ship's galley and adjacent storage compartments. The total weight of all procured materials in the final procurement plan will be recorded as the total weight of procured materials, and the total volume of all procured materials will be recorded as the total volume of procured materials. When the total weight of procured materials exceeds the ship's remaining deadweight, we will sequentially eliminate materials starting with the lowest priority. Each time, we will eliminate the entire quantity of one type of low-priority material, recalculate the total weight of procured materials, and repeat this process until the total weight of procured materials does not exceed the ship's remaining deadweight. For example, if the total weight of the procured materials is 5000 kg, and the ship's remaining deadweight is 4000 kg, and the low-priority materials include 200 kg of flour, 150 kg of sugar, and 100 kg of instant noodles, first remove the least urgent 200 kg of flour from the low-priority list. The total weight becomes 4800 kg, still exceeding 4000 kg. Then remove 150 kg of sugar, resulting in a total weight of 4650 kg, still exceeding 4000 kg. Next, remove 100 kg of instant noodles, bringing the total weight to 4550 kg, still exceeding 4000 kg. If the weight limit is still exceeded after removing all low-priority materials, continue removing medium-priority materials until the deadweight constraint is met. Similarly, when the total volume of the procured materials exceeds the ship's remaining capacity, start removing materials from the low-priority list, eliminating the entire volume of one type of low-priority material at a time, recalculating the total volume of the procured materials, until it does not exceed the ship's remaining capacity. After the removal process, the remaining materials and their procured quantities constitute the material loading list. The list includes the category, weight, volume, and priority level of each item.

[0048] The material categories, weights, and volume parameters from the material loading list are input into the 3D spatial packing simulator. The 3D spatial packing simulator is a pre-built computer program module, and its construction process is as follows: Based on the actual internal dimensions of the ship's galley cold storage and dry cargo warehouse, a 3D coordinate system is established. The origin is at the lower rear corner of the warehouse. The X-axis is along the length of the warehouse (in meters), the Y-axis is along the width of the warehouse (in meters), and the Z-axis is along the height of the warehouse (in meters). The warehouse's length, width, and height are 8 meters, 5 meters, and 3 meters, respectively, with a total volume of 120 cubic meters. The simulator uses a rule-based heuristic algorithm for packing. The rules include: sorting materials from highest to lowest density, placing denser materials at the bottom; density is calculated as the weight of the material divided by its volume (in kilograms per cubic meter); and placing fragile materials such as eggs and bread at the top or away from the hatches. The simulator's workflow is as follows: first, the materials in the material loading list are sorted from highest to lowest density; if the densities are the same, they are sorted from highest to lowest priority. Then, each type of material is taken out sequentially, and standardized packaging box sizes are preset according to the material's volume and dimensions. For example, rice is packaged in a 0.6m x 0.4m x 0.3m cardboard box, with a volume of 0.072 cubic meters. The simulator starts from the bottom of the warehouse and scans available space units layer by layer from left to right and front to back, placing the material in the first space that can accommodate its volume without overlapping with other materials. When placing materials, ensure that the bottom height of denser materials is less than that of less dense materials, i.e., heavier items at the bottom and lighter items on top. For fragile materials, additional constraints are added during placement, prohibiting the placement of any other materials on top of them.

[0049] The simulator records the final placement location of each type of material as storage location coordinates, represented by the coordinates of the lower left corner and the upper right corner. For example, the storage location coordinates for rice are X from 1.0 m to 1.6 m, Y from 0.5 m to 0.9 m, and Z from 0.0 m to 0.3 m. The simulator also generates an unloading sequence number based on the material's storage location height and distance from the hatch. Materials at lower heights and closer to the hatch have smaller unloading sequence numbers, indicating priority for unloading. The sum of all material storage location coordinates and unloading sequence numbers yields a cargo loading map, which is output as a 3D spatial grid.

[0050] The storage location coordinates and unloading sequence number of each material are extracted from the cargo loading diagram to obtain the supply task queue. The supply task queue is sorted in ascending order of unloading sequence number. Each queue entry includes material identifier, storage location coordinates, unloading sequence number, material weight, and material volume. Action coding instructions are generated for unloading actions based on the supply task queue. Each action coding instruction includes action sequence number, material identifier, source hold coordinates, and target exit position. The action sequence number starts from 1 and increments sequentially. The material identifier uses a category identifier. The source hold coordinates are the storage location coordinates of the material in the cargo loading diagram, represented by three coordinate ranges. The target exit position is the location coordinate of the warehouse door. For example, if the warehouse door is located at X = 8 meters, the target exit position is set to X = 8 meters, Y = 2.5 meters, and Z = 1.5 meters. When generating action coding instructions, one instruction is generated for each material in the queue. The instruction format is: sequence number, material identifier, source hold coordinate range X, source hold coordinate range Y, source hold coordinate range Z, target exit coordinate X, target exit coordinate Y, and target exit coordinate Z. For example, the first instruction is "1,R001,1.0-1.6,0.5-0.9,0.0-0.3,8.0,2.5,1.5".

[0051] After summarizing all action code instructions, the simulation executes them and calculates the remaining material inventory and ship center of gravity offset after completion, obtaining the logistics management output. The simulation executes each instruction sequentially according to its action number. Each execution moves the corresponding material from the source hold to the target exit, while simultaneously updating the ship's remaining deadweight and remaining volume. After completion, the remaining material inventory equals the original remaining material inventory plus the weight of the materials actually procured and successfully loaded. The remaining material inventory is calculated separately for each material category. The ship's center of gravity offset is calculated as follows: Before loading, the ship's current center of gravity coordinates are obtained in meters. Each successfully loaded material has its weight and storage location coordinates. The weight of the material is multiplied by the X, Y, and Z values ​​of the storage location coordinates to obtain three torque components. The torque components of all materials are summed and then divided by the total weight of the ship after loading to obtain the new center of gravity coordinates after loading. The center of gravity offset equals the new center of gravity coordinates minus the vector and magnitude of the original center of gravity coordinates. For example, if the original center of gravity coordinates are X = 12.0 meters, Y = 3.0 meters, and Z = 2.5 meters, and the new center of gravity coordinates are X = 12.3 meters, Y = 3.1 meters, and Z = 2.6 meters, then the offset is 0.3 meters in the X direction, 0.1 meters in the Y direction, and 0.1 meters in the Z direction. The total offset is approximately 0.33 meters (√0.3² + 0.1² + 0.1²). The logistics management output includes the final inventory balance for each material, the ship's center of gravity offset, the material loading list, and the action code instruction list. This step ensures that the procurement plan is feasible under the actual load and volume constraints of the ship through a priority elimination mechanism, generates accurate loading diagrams and unloading sequences through three-dimensional spatial container loading simulation, and transforms the loading plan into an executable loading and unloading operation sequence through action coding instructions. The final logistics management output results are directly used to guide actual supply operations, realizing closed-loop management of the entire process from decision-making to execution.

[0052] The second embodiment of the present invention provides an intelligent food inventory management system for a large ship kitchen, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the method described above.

[0053] It should be noted that the intelligent food inventory management system for large ship kitchens provided in this embodiment of the invention is used to execute all the process steps of the intelligent food inventory management method for large ship kitchens described in the above embodiment. The working principles and beneficial effects of the two are one-to-one, so they will not be described again.

[0054] It should be noted that the system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0055] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.

Claims

1. A method for intelligent management of food inventory in the kitchen of large ships, characterized in that, include: By acquiring electrical current, refrigerator temperature, material consumption rate, material inventory balance and remaining sailing time, time-series trend decomposition and prediction are performed to obtain short-term consumption trend values. The ship's speed and significant wave height are obtained, and the short-term consumption trend value is corrected for resistance consumption deviation trend to obtain the corrected consumption trend value. Obtain the original material list of the supply port, conduct differentiated supply and demand matching analysis, and obtain a set of supply information; Based on the corrected consumption trend value and the supply information set, trigger threshold materials are screened to obtain category identifiers. Based on the category identifiers, port capacity assessment based on consumption time is performed to obtain supply urgency level. Based on the supply urgency level, cross-port procurement volume is optimized and allocated to obtain an optimized procurement list. The optimal solution vector is obtained by constructing a minimum number of ports based on the optimized procurement list and the supply information set, and the final procurement plan is generated according to the port procurement instructions based on the optimal solution vector. Based on the final procurement plan, a ship priority capacity constraint loading simulation was performed to obtain the logistics management output results.

2. The intelligent food inventory management method for large ship kitchens according to claim 1, characterized in that, The process involves acquiring electrical current, refrigerator temperature, material consumption rate, remaining material inventory, and remaining sailing time, then performing time-series trend decomposition and prediction to obtain short-term consumption trend values, including: The electrical current, the refrigerator temperature, the material consumption rate, the remaining material inventory, and the remaining sailing time are time-aligned to generate a multi-dimensional synchronization sequence; The multidimensional synchronization sequence is split into periodic components, trend components and residual components using the seasonal trend decomposition method, and the trend components are extracted as consumption feature components. Based on the consumed feature components, a sliding window is used to extract feature data within a preset window to construct a time window feature matrix; Based on the time window feature matrix, input the data into the Long Short-Term Memory network for state memory and normalized prediction, and output the intermediate prediction value. Based on the predicted intermediate value, it is denormalized to restore the predicted consumption value, and then superimposed with the preset baseline consumption value to obtain the short-term consumption trend value.

3. The intelligent food inventory management method for large ship kitchens according to claim 1, characterized in that, The process of acquiring ship speed and significant wave height, and correcting the short-term consumption trend value for resistance consumption deviation to obtain a corrected consumption trend value, includes: The ship's resistance variables are calculated by inputting the ship's speed and effective wave height into the ship resistance increment model. Divide the resistance variable by the preset standard resistance to obtain the consumption rate correction coefficient; Multiply the consumption rate correction factor by the material consumption rate to obtain the consumption deviation value; The consumption deviation value is superimposed on the short-term consumption trend value to obtain the corrected consumption trend value.

4. The intelligent food inventory management method for large ship kitchens according to claim 1, characterized in that, The original material list of the supply port is obtained and subjected to differentiated supply and demand matching analysis to obtain a set of supply information, including: Obtain the original material list of the supply port, identify the name differences and substitutes in the original material list, and generate a differentiated material catalog; Based on the differentiated material catalog, the upper limit of the supply capacity of the supply port and the time required for cargo transfer are obtained, thus obtaining the set of port supply capacity. Multiplying the resource consumption rate by the remaining sailing time yields the estimated total consumption. Subtracting the current inventory balance from the estimated total consumption gives the resource shortage figure. Based on the port supply capacity set and the material shortage value, the available material quantity for supplying the port is obtained, and a supply information set containing category differences and supply capacity information is generated.

5. The intelligent food inventory management method for large ship kitchens according to claim 4, characterized in that, The process involves screening materials based on the corrected consumption trend value and the supply information set to obtain category identifiers, assessing port capacity based on consumption duration using these category identifiers to determine supply urgency levels, and optimizing cross-port procurement allocation based on these supply urgency levels to obtain an optimized procurement list, including: Extract the category identifier of the material corresponding to the corrected consumption trend value being greater than the preset supply trigger threshold; Based on the category identifier, retrieve the port availability of the corresponding material at the supply port and the time required for the transfer from the supply information set; Divide the remaining inventory of materials by the corrected consumption trend value to obtain the remaining available days, and determine the supply urgency level according to the remaining available days in ascending order; Based on the supply urgency level and the available quantity at the port, the materials with the highest urgency are prioritized and allocated to the supply port with the shortest relocation time. The procurement volume of each supply port is then allocated according to the material priority to generate an optimized procurement list.

6. The intelligent food inventory management method for large ship kitchens according to claim 5, characterized in that, The process of constructing and solving the optimal solution vector by planning for the minimum number of ports based on the optimized procurement list and the supply information set, and generating the final procurement plan according to the port procurement instructions based on the optimal solution vector, includes: Based on the optimized procurement list and the supply information set, a Boolean variable is introduced for the supply port to indicate whether to perform procurement at the supply port. With the goal of minimizing the sum of all the Boolean variables, the procurement quantity of materials at the supply port is used as a continuous decision variable to construct a linear programming model. The linear programming model is input into a preset mixed-integer linear programming solver for iterative solution to obtain the optimal solution vector for replenishing port procurement volume; Based on the optimal solution vector, independent procurement instructions are generated according to the allocation of supply ports, and a final procurement plan is determined, including the procurement port, the category of procurement materials, the weight of procurement materials, the total volume of procurement materials, and the total weight of procurement materials.

7. The intelligent food inventory management method for large ship kitchens according to claim 6, characterized in that, The constraints of the linear programming model include: If the procurement volume at the supply port is greater than zero, it must not be less than the preset minimum order quantity and must not exceed the upper limit of the supply capacity. The sum of the procurement volumes at all supply ports equals the material shortage value.

8. The intelligent food inventory management method for large ship kitchens according to claim 6, characterized in that, The simulation of ship priority capacity constraint loading based on the final procurement plan, and the resulting logistics management output, include: Based on the final procurement plan, cargo space priorities are assigned to the procured materials. When the total weight of the purchased materials exceeds the pre-obtained remaining deadweight of the ship, materials with lower priority are eliminated sequentially until the total weight of the purchased materials does not exceed the remaining deadweight of the ship. When the total volume of the purchased materials exceeds the pre-obtained remaining volume of the ship, materials with lower priority are eliminated sequentially until the total volume of the purchased materials does not exceed the remaining volume of the ship, thus obtaining the material loading list. Input the material category, weight and volume parameters in the material loading list into the three-dimensional space packing simulator, and generate a cargo loading diagram according to the rule of placing dense materials at the bottom and fragile materials at the top. Extract the storage location coordinates and unloading sequence number of each material from the cargo loading diagram to obtain the supply task queue, and generate action coding instructions containing action sequence number, material identifier, source hold coordinates and target exit position for the unloading action based on the supply task queue. After summarizing all the action coding instructions, simulate their execution and calculate the remaining material inventory and the ship's center of gravity offset after execution to obtain the logistics management output results.

9. A smart food inventory management system for a large ship's kitchen, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the method according to any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the program implements the steps of the method described in any one of claims 1 to 8.

Images