Nuclear power plant spare part demand prediction method and device based on Bayesian reasoning

Abstract

The invention particularly relates to a nuclear power plant spare part demand prediction method and device based on Bayesian reasoning, and belongs to the technical field of spare part management. The method comprises the following steps: collecting nuclear power plant spare part demand original data; preprocessing the nuclear power plant spare part demand original data to obtain nuclear power plant spare part demand data; constructing a Bayesian reasoning model; calculating the spare part demand rate in the spare part prediction period by using a Bayesian inference model; and according to the spare part demand rate in the spare part prediction period, optimizing the inventory ordering quantity and the inventory ordering cost in the spare part prediction period. The device is used for implementing the steps of the method. The accuracy of nuclear power plant spare part demand prediction is improved, a basis is provided for decision making of production or purchase plans, and therefore the efficiency and accuracy of nuclear power plant spare part inventory management are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of spare parts management, and in particular to a method and device for predicting spare parts demand of a nuclear power plant based on Bayesian reasoning. Background Art

[0002] Spare parts are the material foundation for nuclear power plant equipment maintenance. However, premature delivery of some spare parts leads to excessive spare parts inventory, increasing nuclear power plant operating costs. Late delivery of some spare parts leads to low spare parts inventory, preventing on-site demand from being met and impacting spare parts supply availability. The primary task of spare parts inventory management is to resolve the conflict between spare parts supply availability and spare parts inventory costs. To strike a balance between spare parts supply availability and spare parts inventory costs, maintaining a reasonable spare parts inventory is essential. Maintaining a reasonable spare parts inventory requires that spare parts arrive on demand, making spare parts inventory forecasting and management crucial.

[0003] Traditional spare parts demand forecasting methods are typically based on deterministic or empirical models, such as the economic lot size model, the safety stock model, and the news vendor model. While simple and easy to use, these methods often overlook the uncertainty and dynamic nature of spare parts demand, leading to inaccurate forecasts. This, in turn, impacts the efficiency and accuracy of spare parts inventory management, increases spare parts inventory costs, and increases the risk of stockouts. Therefore, leveraging advanced mathematical models and algorithms to improve the accuracy of spare parts demand forecasts has become a pressing issue. Summary of the Invention

[0004] Based on this, it is necessary to provide a nuclear power plant spare parts demand prediction method and device based on Bayesian reasoning to improve the accuracy of nuclear power plant spare parts demand prediction, provide a basis for decision-making on production or procurement plans, and thus improve the efficiency and accuracy of nuclear power plant spare parts inventory management.

[0005] In order to solve the above technical problems, on the one hand, the present invention provides a method for predicting the demand for spare parts of a nuclear power plant based on Bayesian reasoning, comprising the following steps: Step 101: Collect original data on spare parts demand of nuclear power plants; pre-process the original data on spare parts demand of nuclear power plants to obtain spare parts demand data of nuclear power plants; Step 102: construct a Bayesian reasoning model; Step 103: Calculate the spare parts demand rate within the spare parts forecast period using the Bayesian inference model; Step 104: Optimize the inventory order quantity and inventory order cost within the spare parts forecast period according to the spare parts demand rate within the spare parts forecast period.

[0006] As one possible implementation method, a piece of original spare parts demand data is generated within a spare parts observation cycle. The original spare parts demand data includes original historical spare parts demand data and original real-time spare parts demand data. The spare parts observation cycle is the length of the time window for collecting original real-time spare parts demand data. Spare parts demand data includes historical spare parts demand data and real-time spare parts demand data; the original historical spare parts demand data is preprocessed to obtain the historical spare parts demand data; the original real-time spare parts demand data is preprocessed to obtain the real-time spare parts demand data; The original historical data of spare parts demand, the original real-time data of spare parts demand, the historical data of spare parts demand, and the real-time data of spare parts demand are all vectors containing the following features: spare part type, spare part classification, spare part usage frequency, spare part inventory, spare part usage location, spare part maintenance plan, spare part certificate, spare part entry time, and spare part exit time. Among them, spare parts types include wearing parts and non-wearing parts; spare parts classification includes four levels: critical sensitive, critical, important, and general; spare parts usage frequency is the interval between two spare parts issuance records or typical preventive periodic outbound delivery records; spare parts inventory is the current spare parts inventory; spare parts usage locations include defect repair, preventive maintenance and quality defect reporting; spare parts maintenance plans include disassembly inspection and regular replacement; spare parts vouchers include spare parts allocation, spare parts transfer and spare parts consumption.

[0007] As one of the feasible methods, the original spare parts demand data is preprocessed to obtain the spare parts demand data, including the following steps: removing the original spare parts demand data that is not related to spare parts consumption, and then filling the missing values of the features in the original spare parts demand data to obtain the spare parts demand data; Among them, the original data of spare parts demand that is not related to spare parts consumption is eliminated, including the original data of spare parts demand with spare parts vouchers being spare parts transfer or spare parts warehouse transfer. Filling missing values of features in the original spare parts demand data includes the following steps: The original data of spare parts demand is divided into complete samples and missing samples. In the complete samples, the features do not have missing values, while in the missing samples, the features have missing values. The features with missing values in the missing samples are called missing features. The features in the missing samples and complete samples are divided into numerical and categorical features. The categorical features include spare part type, spare part classification, spare part usage location, spare part maintenance plan, and spare part voucher. The numerical features include spare part usage frequency, spare parts inventory, spare parts entry time, and spare parts exit time. The comprehensive distance between the missing sample and each complete sample is calculated separately. Sort each complete sample in ascending order according to the comprehensive distance between the missing sample and the complete sample, and select the first A complete sample; When the missing feature is numerical, The weighted average of the corresponding feature values in the complete samples is used to fill in the missing feature values in the missing samples; When the missing feature is of categorical type, The mode of the corresponding feature values in the complete samples is taken as the value of the missing feature in the missing sample to fill in the missing feature.

[0008] As one of the feasible methods, calculating the comprehensive distance between missing samples and complete samples includes the following steps: Calculate the distance between the missing sample and the numerical feature in the complete sample; Calculate the distance between the categorical features in the missing samples and the complete samples; The distance between the missing sample and the complete sample is weighted averaged by the distance between the numerical features and the categorical features to obtain the comprehensive distance between the missing sample and the complete sample.

[0009] As one of the possible ways to do this, we can calculate the distance between the missing sample and the numerical feature in the complete sample, which includes the following steps: When the value of a numerical feature is not missing in both the missing sample and the complete sample, the numerical feature is called a valid numerical feature. To determine the valid numerical feature index set of missing samples and complete samples: Calculate the Minkowski distance between the missing and complete samples for numerical features: Normalize the Minkowski distance between the missing sample and the numerical feature in the complete sample to obtain the distance between the missing sample and the numerical feature in the complete sample: in, is a valid numerical feature index set of missing samples and complete samples, is the Minkowski distance between the numerical features in the missing sample and the complete sample, is the distance between the missing sample and the numerical feature in the complete sample, is the total number of valid numerical feature dimensions in missing samples and complete samples; is a valid numerical feature dimension index, For missing samples, For the complete sample, The missing sample Valid numerical features, For the complete sample Valid numerical features, is the distance order.

[0010] As one of the feasible methods, the distance between the missing samples and the classification features in the complete samples is calculated, which includes the following steps: When the value of a categorical feature does not exist in both missing samples and complete samples, the categorical feature is called a valid categorical feature; Determine the valid classification feature index set for missing samples and complete samples: Calculate the distance between the missing samples and the classification features in the complete samples: in, is the effective classification feature index set of missing samples and complete samples, is the difference between the classification characteristics in the missing sample and the complete sample, is the distance between the classification characteristics in the missing sample and the complete sample; is the effective classification feature dimension index, For missing samples, For the complete sample, The missing sample Valid classification features, For the complete sample Valid classification features, is the total number of valid classification feature dimensions in missing samples and complete samples.

[0011] As one of the feasible methods, the distance between the missing sample and the complete sample in terms of numerical features and the distance between the categorical features is weighted averaged according to the following formula to obtain the comprehensive distance between the missing sample and the complete sample: in, is the distance weight between the missing sample and the numerical feature in the complete sample, is the distance weight between the classification features in the missing sample and the complete sample, .

[0012] As one possible implementation method, step 102, constructing a Bayesian inference model, includes the following steps: Step 1021: Construct a priori Gamma probability density function to describe the prior distribution of spare parts demand rate: Step 1022: Construct a composite Gamma-Poisson probability function to calculate the spare parts replenishment lead time. Internal Probability of spare parts demand: Step 1023: Construct a basic spare parts inventory constraint equation to solve the minimum basic spare parts inventory that meets the spare parts service level: Step 1024: Construct a posterior Gamma probability density function to dynamically update the parameters in the Gamma probability density function. and : Step 1025: Construct a posterior confidence interval equation to predict the upper confidence limit of the spare parts demand rate that meets the spare parts service level: in, is the prior Gamma probability density function; is the shape parameter, is the scale parameter is the prior gamma function, ; is the posterior Gamma probability density function; is the updated shape parameter, is the updated scale parameter is the posterior gamma function, ; is the spare parts observation period, is the number of spare parts requirements within the spare parts observation period; is the spare parts demand rate, which represents the average number of spare parts demands per unit time; k is the number of spare parts demands; L is the spare parts replenishment lead time, which represents the time interval from the initiation of the spare parts order to the arrival of the spare parts in the warehouse, and is set based on supplier data or procurement data; is the average number of requests during the spare parts replenishment lead time; is the spare parts service level, which represents the probability threshold that the spare parts inventory meets the spare parts demand, set according to the spare parts management strategy; s is the spare parts basic inventory, which covers the spare parts replenishment lead time The probability of spare parts demand at least once within s-1 is not less than ; The upper confidence limit for the spare parts demand rate that meets the spare parts service level.

[0013] As one possible implementation method, step 103, using a Bayesian inference model to calculate the spare parts demand rate within the spare parts forecast period, includes the following steps: Step 1031: Calculate the parameters in the prior Gamma probability density function based on the historical data of spare parts demand. and ; Step 1032: The calculated and Substitute the prior Gamma probability density function to obtain the prior Gamma probability density function; Step 1033: Set the spare parts replenishment lead time L and spare parts service level Substitute into the spare parts basic inventory constraint equation to solve the minimum spare parts basic inventory that meets the spare parts service level ; Step 1034: Dynamically update the prior Gamma probability density function based on the real-time data of spare parts demand. and , obtain the posterior Gamma probability density function, and substitute the posterior Gamma probability density function into the posterior confidence interval equation to calculate the upper confidence limit of the spare parts demand rate that meets the spare parts service level , as the spare parts demand rate within the spare parts forecast period; The spare parts forecast cycle is the length of the time window for forecasting future spare parts demand and is set according to the spare parts management strategy.

[0014] As one possible implementation method, step 104, optimizing the inventory order quantity and inventory order cost within the spare parts forecast period according to the spare parts demand rate within the spare parts forecast period, includes the following steps: Step 1041: Obtain spare parts inventory ordering data, which includes fixed cost of a single spare parts order, cost of a single spare parts order, current spare parts inventory, current spare parts in transit inventory, and spare parts forecast cycle. Step 1042: The target inventory within the spare parts forecast period is obtained by multiplying the spare parts demand rate within the spare parts forecast period by the spare parts forecast period. Step 1042: Calculate the optimal inventory order quantity within the spare parts forecast period and determine the minimum inventory order cost within the spare parts forecast period: in, is the optimal inventory order quantity within the spare parts forecast period, is the minimum inventory ordering cost within the spare parts forecast period, Fixed cost for a single order of spare parts, is the single ordering cost of spare parts, is the forecast period, is the target inventory within the spare parts forecast period, is the current inventory of spare parts; N is the current in-transit inventory of spare parts, indicating the number of spare parts that have been ordered but not yet arrived.

[0015] On the other hand, the present invention also provides a nuclear power plant spare parts demand prediction device based on Bayesian reasoning, including a memory and a processor, wherein the memory stores computer-readable instructions, and when the processor executes the computer-readable instructions, it implements the steps of the above-mentioned nuclear power plant spare parts demand prediction method based on Bayesian reasoning.

[0016] Beneficial technical effects of the present invention: The method and device for predicting spare parts demand for nuclear power plants based on Bayesian reasoning of the present invention establish a Bayesian reasoning model by analyzing historical spare parts demand data, while fully considering the uncertainty and dynamic nature of spare parts demand, and calculating the probability distribution of spare parts demand in combination with the Bayesian reasoning model. The spare parts demand prediction results are updated and corrected by the Bayesian reasoning model, which can more accurately predict spare parts demand, thereby formulating a more reasonable spare parts inventory management strategy and improving the efficiency and accuracy of spare parts inventory management. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 The flowchart of an embodiment of the method for predicting spare parts demand of nuclear power plants based on Bayesian reasoning of the present invention is shown. DETAILED DESCRIPTION

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application belongs; the terms used in the specification of the application are only for the purpose of describing specific embodiments and are not intended to limit this application; the terms "including" and "having" and any variations thereof in the specification and claims of this application and the above-mentioned figure descriptions are intended to cover non-exclusive inclusions.

[0019] References herein to "embodiments" mean that a particular feature, structure, or characteristic described in connection with the embodiments may be included in at least one embodiment of the present application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute an independent or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0020] The technical solution of the present invention is described clearly and completely below with reference to the accompanying drawings and specific embodiments.

[0021] refer to Figure 1 , shows a flow chart of an embodiment of a method for predicting spare parts demand for a nuclear power plant based on Bayesian reasoning, the method for predicting spare parts demand for a nuclear power plant comprising the following steps: Step 101: Collect original data on spare parts demand of nuclear power plants; pre-process the original data on spare parts demand of nuclear power plants to obtain spare parts demand data of nuclear power plants; Step 102: Building a Bayesian inference model based on the nuclear power plant spare parts demand data; Step 103: Calculate the spare parts demand rate within the spare parts forecast period using the Bayesian inference model; Step 104: Optimize the inventory order quantity and inventory order cost within the spare parts forecast period according to the spare parts demand rate within the spare parts forecast period.

[0022] Spare parts for nuclear power plants are typically high-value and slow-moving. In this embodiment, as one possible implementation, a piece of raw spare parts demand data is generated within a spare parts observation period. The raw spare parts demand data includes raw historical spare parts demand data and raw real-time spare parts demand data. The spare parts observation period is the time window length for collecting raw real-time spare parts demand data. Spare parts demand data includes historical spare parts demand data and real-time spare parts demand data; the original historical spare parts demand data is preprocessed to obtain the historical spare parts demand data; the original real-time spare parts demand data is preprocessed to obtain the real-time spare parts demand data; The original historical data of spare parts demand, the original real-time data of spare parts demand, the historical data of spare parts demand, and the real-time data of spare parts demand are all vectors containing the following features: spare part type, spare part classification, spare part usage frequency, spare part inventory, spare part usage location, spare part maintenance plan, spare part certificate, spare part entry time, and spare part exit time. Among them, spare parts types include wearing parts and non-wearing parts, which are used to judge whether the spare parts usage is reasonable; Spare parts classification is defined according to equipment classification, mainly considering the function of spare parts, including four levels: critical sensitive, critical, important, and general; The frequency of spare parts use is the interval between two spare parts issuance records or the typical preventive periodic delivery records; the spare parts inventory is the current inventory of spare parts; The spare parts usage location is a virtual usage location, which is mainly used to clarify the purpose of the spare parts, including defect repair, preventive maintenance and quality defect reporting; The spare parts maintenance plan is the maintenance plan corresponding to the use of spare parts, including disassembly inspection and regular replacement; Spare parts vouchers include spare parts transfer, spare parts transfer and spare parts consumption. Spare parts transfer and spare parts transfer are not related to spare parts demand.

[0023] The original spare parts demand data may be irrelevant to the actual consumption of spare parts, and features in the original spare parts demand data may contain missing values due to the standardization of records. In this embodiment, as one of the possible implementation methods, the original spare parts demand data is preprocessed to obtain the spare parts demand data, including the following steps: removing the original spare parts demand data that is irrelevant to spare parts consumption, and then filling the missing values of features in the original spare parts demand data to obtain the spare parts demand data; Among them, the original data of spare parts demand that is not related to spare parts consumption is eliminated, including the original data of spare parts demand with spare parts vouchers being spare parts transfer or spare parts warehouse transfer. Filling missing values of features in the original spare parts demand data includes the following steps: The original data of spare parts demand is divided into complete samples and missing samples. In the complete samples, the features do not have missing values, while in the missing samples, the features have missing values. The features with missing values in the missing samples are called missing features. The features in the missing samples and complete samples are divided into numerical and categorical features. The categorical features include spare part type, spare part classification, spare part usage location, spare part maintenance plan, and spare part voucher. The numerical features include spare part usage frequency, spare parts inventory, spare parts entry time, and spare parts exit time. The comprehensive distance between the missing sample and each complete sample is calculated separately. Sort each complete sample in ascending order according to the comprehensive distance between the missing sample and the complete sample, and select the first A complete sample; When the missing feature is numerical, The weighted average of the corresponding feature values in the complete samples is used to fill in the missing feature values in the missing samples; When the missing feature is of categorical type, The mode of the corresponding feature values in the complete samples is taken as the value of the missing feature in the missing sample to fill in the missing feature.

[0024] In this embodiment, as one of the achievable methods, calculating the comprehensive distance between missing samples and complete samples includes the following steps: Calculate the distance between the missing sample and the numerical feature in the complete sample; Calculate the distance between the categorical features in the missing samples and the complete samples; The distance between the missing sample and the complete sample is weighted averaged by the distance between the numerical features and the categorical features to obtain the comprehensive distance between the missing sample and the complete sample.

[0025] In this embodiment, as one of the possible implementation methods, calculating the distance between the missing sample and the numerical feature in the complete sample includes the following steps: When the value of a numerical feature is not missing in both the missing sample and the complete sample, the numerical feature is called a valid numerical feature. To determine the valid numerical feature index set of missing samples and complete samples: Calculate the Minkowski distance between the missing and complete samples for numerical features: Normalize the Minkowski distance between the missing sample and the numerical feature in the complete sample to obtain the distance between the missing sample and the numerical feature in the complete sample: in, is a valid numerical feature index set of missing samples and complete samples, is the Minkowski distance between the numerical features in the missing sample and the complete sample, is the distance between the missing sample and the numerical feature in the complete sample, is the total number of valid numerical feature dimensions in missing samples and complete samples; is a valid numerical feature dimension index, For missing samples, For the complete sample, The missing sample Valid numerical features, For the complete sample Valid numerical features, is the distance order.

[0026] In this embodiment, as one of the possible implementation methods, =2, =5.

[0027] In this embodiment, as one of the possible implementation methods, calculating the distance between the classification features in the missing sample and the complete sample includes the following steps: When the value of a categorical feature does not exist in both missing samples and complete samples, the categorical feature is called a valid categorical feature; Determine the valid classification feature index set for missing samples and complete samples: Calculate the distance between the missing samples and the classification features in the complete samples: in, is the effective classification feature index set of missing samples and complete samples, is the difference between the classification characteristics in the missing sample and the complete sample, is the distance between the classification characteristics in the missing sample and the complete sample; is the effective classification feature dimension index, For missing samples, For the complete sample, The missing sample Valid classification features, For the complete sample Valid classification features, is the total number of valid classification feature dimensions in missing samples and complete samples.

[0028] In this embodiment, as one of the possible implementation methods, the distance between the missing sample and the complete sample in terms of numerical features and the distance between the categorical features is weighted averaged according to the following formula to obtain the comprehensive distance between the missing sample and the complete sample: in, is the distance weight between the missing sample and the numerical feature in the complete sample, is the distance weight between the classification features in the missing sample and the complete sample, .

[0029] In this embodiment, as one of the possible implementation methods, when the impact of the numerical feature on the spare parts demand is less than the impact of the categorical feature on the spare parts demand, it is set Less than When the impact of numerical features on spare parts demand is greater than that of categorical features on spare parts demand, set Greater than When the impact of numerical features on spare parts demand is equal to the impact of categorical features on spare parts demand, let = .

[0030] In this embodiment, as one possible implementation method, step 102, constructing a Bayesian inference model, includes the following steps: Step 1021: Construct a priori Gamma probability density function to describe the prior distribution of spare parts demand rate: Step 1022: Construct a composite Gamma-Poisson probability function to calculate the spare parts replenishment lead time. Internal Probability of spare parts demand: Step 1023: Construct a basic spare parts inventory constraint equation to solve the minimum basic spare parts inventory that meets the spare parts service level: Step 1024: Construct a posterior Gamma probability density function to dynamically update the parameters in the Gamma probability density function. and : Step 1025: Construct a posterior confidence interval equation to predict the upper confidence limit of the spare parts demand rate that meets the spare parts service level: in, is the prior Gamma probability density function; is the shape parameter, is the scale parameter is the prior gamma function, ; is the posterior Gamma probability density function; is the updated shape parameter, is the updated scale parameter is the posterior gamma function, ; is the spare parts observation period, is the number of spare parts requirements within the spare parts observation period; is the spare parts demand rate, which represents the average number of spare parts demands per unit time; k is the number of spare parts demands; L is the spare parts replenishment lead time, which represents the time interval from the initiation of the spare parts order to the arrival of the spare parts in the warehouse, and is set based on supplier data or procurement data; is the average number of requests during the spare parts replenishment lead time; is the spare parts service level, which represents the probability threshold that the spare parts inventory meets the spare parts demand, set according to the spare parts management strategy; s is the spare parts basic inventory, which covers the spare parts replenishment lead time The probability of spare parts demand at least once within s-1 is not less than ; The upper confidence limit for the spare parts demand rate that meets the spare parts service level.

[0031] In this embodiment, as one possible implementation method, step 103, using a Bayesian inference model to calculate the spare parts demand rate within the spare parts forecast period, includes the following steps: Step 1031: Calculate the parameters in the prior Gamma probability density function based on the historical data of spare parts demand. and ; Step 1032: The calculated and Substitute the prior Gamma probability density function to obtain the prior Gamma probability density function; Step 1033: Set the spare parts replenishment lead time L and spare parts service level Substitute into the spare parts basic inventory constraint equation to solve the minimum spare parts basic inventory that meets the spare parts service level ; Step 1034: Dynamically update the prior Gamma probability density function based on the real-time data of spare parts demand. and , obtain the posterior Gamma probability density function, and substitute the posterior Gamma probability density function into the posterior confidence interval equation to calculate the upper confidence limit of the spare parts demand rate that meets the spare parts service level , as the spare parts demand rate within the spare parts forecast period; The spare parts forecast cycle is the length of the time window for forecasting future spare parts demand and is set according to the spare parts management strategy.

[0032] In this embodiment, as one of the possible implementation methods, step 1031 is to calculate the parameters in the prior Gamma probability density function based on the historical data of spare parts demand. and , including the following steps: For each piece of spare parts demand historical data, calculate the corresponding spare parts demand rate; Calculate the spare parts demand rate of each spare parts demand historical data ; according to ,calculate and ; , .

[0033] In this embodiment, as one possible implementation, step 104, optimizing the inventory order quantity and inventory order cost within the spare parts forecast period based on the spare parts demand rate within the spare parts forecast period, includes the following steps: Step 1041: Obtain spare parts inventory ordering data, which includes fixed cost of a single spare parts order, cost of a single spare parts order, current spare parts inventory, current spare parts in transit inventory, and spare parts forecast cycle. Step 1042: The target inventory within the spare parts forecast period is obtained by multiplying the spare parts demand rate within the spare parts forecast period by the spare parts forecast period. Step 1042: Calculate the optimal inventory order quantity within the spare parts forecast period and determine the minimum inventory order cost within the spare parts forecast period: in, is the optimal inventory order quantity within the spare parts forecast period, is the minimum inventory ordering cost within the spare parts forecast period, Fixed cost for a single order of spare parts, is the single ordering cost of spare parts, is the forecast period, is the target inventory within the spare parts forecast period, is the current inventory of spare parts; N is the current in-transit inventory of spare parts, indicating the number of spare parts that have been ordered but not yet arrived.

[0034] As an implementation of the above method, the present invention provides an embodiment of a device for predicting spare parts demand of a nuclear power plant based on Bayesian reasoning. The embodiment of the device corresponds to the embodiment of the method for predicting spare parts demand of a nuclear power plant based on Bayesian reasoning.

[0035] The device described in this embodiment includes a memory, a processor, and a network interface that are interconnected and communicated via a system bus. It should be noted that the embodiment only shows a device having a memory, a processor, and a network interface, but it should be understood that it is not required to implement all the components shown, and more or fewer components can be implemented instead. Among them, those skilled in the art will understand that the device here is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to a microprocessor, an application-specific integrated circuit, a programmable gate array, a digital processor, an embedded device, etc.

[0036] The device can be a computing device such as a desktop computer, notebook, PDA, cloud server, etc. The device can interact with the user through a keyboard, mouse, remote control, touchpad, or voice control device.

[0037] The memory includes at least one type of readable storage medium, including flash memory, hard disk, multimedia card, card-type memory, random access memory, static random access memory, read-only memory, electrically erasable programmable read-only memory, programmable read-only memory, magnetic storage, magnetic disk, optical disk, etc. In some embodiments, the memory may be an internal storage unit of the device, such as the device's hard disk or internal memory. In other embodiments, the memory may also be an external storage device of the device, such as a plug-in hard disk, smart memory card, secure digital card, flash memory card, etc. Of course, the memory may also include both the device's internal storage unit and its external storage device. In this embodiment, the memory is typically used to store the operating system and various application software installed on the device, such as the computer-readable instructions for the aforementioned Bayesian reasoning-based nuclear power plant spare parts demand forecasting method. Furthermore, the memory may also be used to temporarily store various types of data that have been output or will be output.

[0038] In some embodiments, the processor may be a central processing unit, a controller, a microcontroller, a microprocessor, or other data processing chip. The processor is typically used to control the overall operation of the device. In this embodiment, the processor is used to execute computer-readable instructions stored in the memory or process data, such as computer-readable instructions for executing the aforementioned nuclear power plant spare parts demand forecasting method.

[0039] The network interface may include a wireless network interface or a wired network interface, which is generally used to establish a communication connection between the device and other electronic devices.

[0040] The above-described embodiments merely illustrate several implementations of the present invention, and while their descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art would be able to make numerous variations and improvements without departing from the spirit of the present invention, all of which fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be determined by the appended claims.

Claims

1. A method for predicting spare parts demand for nuclear power plants based on Bayesian reasoning, characterized in that: The following steps are involved: Step 101: Collect original data on spare parts demand of nuclear power plants; pre-process the original data on spare parts demand of nuclear power plants to obtain spare parts demand data of nuclear power plants; Step 102: construct a Bayesian inference model; Step 103: Calculate the spare parts demand rate within the spare parts forecast period using the Bayesian inference model; Step 104: Optimize the inventory order quantity and inventory order cost within the spare parts forecast period according to the spare parts demand rate within the spare parts forecast period.

2. The method for predicting nuclear power plant spare parts demand based on Bayesian reasoning according to claim 1, characterized in that: During a spare parts observation cycle, a piece of spare parts demand original data is generated. The spare parts demand original data includes the spare parts demand original historical data and the spare parts demand original real-time data. The spare parts observation cycle is the time window length for collecting spare parts demand original real-time data. The spare parts demand data includes historical spare parts demand data and real-time spare parts demand data. The original historical spare parts demand data is preprocessed to obtain the historical spare parts demand data. The original real-time spare parts demand data is preprocessed to obtain the real-time spare parts demand data.

3. The method for predicting nuclear power plant spare parts demand based on Bayesian reasoning according to claim 2, characterized in that: Preprocessing the original spare parts demand data to obtain spare parts demand data includes the following steps: Eliminate the original spare parts demand data that is not related to spare parts consumption, and then fill in the missing values of the features in the original spare parts demand data to obtain the spare parts demand data; Filling missing values of features in the original spare parts demand data includes the following steps: The original data of spare parts demand is divided into complete samples and missing samples. In the complete samples, the features do not have missing values, while in the missing samples, the features have missing values. The features with missing values in the missing samples are called missing features. The features in the missing samples and complete samples are divided into numerical and categorical types, and the comprehensive distance between the missing samples and each complete sample is calculated respectively; Sort each complete sample in ascending order according to the comprehensive distance between the missing sample and the complete sample, and select the first A complete sample; When the missing feature is numerical, The weighted average of the corresponding feature values in the complete samples is used to fill in the missing feature values in the missing samples; When the missing feature is of categorical type, The mode of the corresponding feature values in the complete samples is taken as the value of the missing feature in the missing sample to fill in the missing feature.

4. The method for predicting nuclear power plant spare parts demand based on Bayesian reasoning according to claim 3, characterized in that: Calculating the comprehensive distance between missing samples and complete samples includes the following steps: Calculate the distance between the missing sample and the numerical feature in the complete sample; Calculate the distance between the categorical features in the missing samples and the complete samples; The distance between the missing sample and the complete sample is weighted averaged by the distance between the numerical features and the categorical features to obtain the comprehensive distance between the missing sample and the complete sample.

5. The method for predicting nuclear power plant spare parts demand based on Bayesian reasoning according to claim 4, characterized in that: Calculating the distance between the missing sample and the numerical feature in the complete sample includes the following steps: When the value of a numerical feature is not missing in both the missing sample and the complete sample, the numerical feature is called a valid numerical feature. To determine the valid numerical feature index set of missing samples and complete samples: Calculate the Minkowski distance between the missing and complete samples for numerical features: Normalize the Minkowski distance between the missing sample and the numerical feature in the complete sample to obtain the distance between the missing sample and the numerical feature in the complete sample: in, is a valid numerical feature index set of missing samples and complete samples, is the Minkowski distance between the numerical features in the missing sample and the complete sample, is the distance between the missing sample and the numerical feature in the complete sample, is the total number of valid numerical feature dimensions in missing samples and complete samples; is a valid numerical feature dimension index, For missing samples, For the complete sample, The missing sample Valid numerical features, For the complete sample Valid numerical features, is the distance order.

6. The method for predicting nuclear power plant spare parts demand based on Bayesian reasoning according to claim 4, characterized in that: Calculating the distance between the missing samples and the classification features in the complete samples includes the following steps: When the value of a categorical feature does not exist in both missing samples and complete samples, the categorical feature is called a valid categorical feature; Determine the valid classification feature index set for missing samples and complete samples: Calculate the distance between the missing samples and the classification features in the complete samples: in, is the effective classification feature index set of missing samples and complete samples, is the difference between the classification characteristics in the missing sample and the complete sample, is the distance between the classification characteristics in the missing sample and the complete sample; is the effective classification feature dimension index, For missing samples, For the complete sample, The missing sample Valid classification features, For the complete sample Valid classification features, is the total number of valid classification feature dimensions in missing samples and complete samples.

7. The method for predicting nuclear power plant spare parts demand based on Bayesian reasoning according to claim 4, characterized in that: The distance between the missing sample and the complete sample is weighted averaged by the distance between the numerical features and the categorical features according to the following formula to obtain the comprehensive distance between the missing sample and the complete sample: in, is the distance weight between the missing sample and the numerical feature in the complete sample, is the distance weight between the classification features in the missing sample and the complete sample, .

8. The method for predicting nuclear power plant spare parts demand based on Bayesian reasoning according to claim 1, characterized in that: Step 102, constructing a Bayesian inference model, includes the following steps: Step 1021: Construct a priori Gamma probability density function to describe the prior distribution of spare parts demand rate; Step 1022: Construct a composite Gamma-Poisson probability function to calculate the spare parts replenishment lead time. Internal The probability of secondary spare parts demand; Step 1023: Construct a basic spare parts inventory constraint equation to solve the minimum basic spare parts inventory that meets the spare parts service level; Step 1024: Construct a posterior Gamma probability density function to dynamically update the parameters in the Gamma probability density function. and ; Step 1025: Construct a posterior confidence interval equation to predict the upper confidence limit of the spare parts demand rate that meets the spare parts service level.

9. The method for predicting nuclear power plant spare parts demand based on Bayesian reasoning according to claim 1, characterized in that: Step 103, using the Bayesian inference model to calculate the spare parts demand rate within the spare parts forecast period, includes the following steps: Step 1031: Calculate the parameters in the prior Gamma probability density function based on the historical data of spare parts demand. and ; Step 1032: The calculated and Substitute the prior Gamma probability density function to obtain the prior Gamma probability density function; Step 1033: Set the spare parts replenishment lead time L and spare parts service level Substitute into the spare parts basic inventory constraint equation to solve the minimum spare parts basic inventory that meets the spare parts service level ; Step 1034: Dynamically update the prior Gamma probability density function based on the real-time data of spare parts demand. and , obtain the posterior Gamma probability density function, and substitute the posterior Gamma probability density function into the posterior confidence interval equation to calculate the upper confidence limit of the spare parts demand rate that meets the spare parts service level , as the spare parts demand rate during the spare parts forecast period.

10. The method for predicting nuclear power plant spare parts demand based on Bayesian reasoning according to claim 1, characterized in that: Step 104, optimizing the inventory order quantity and inventory order cost within the spare parts forecast period according to the spare parts demand rate within the spare parts forecast period, includes the following steps: Step 1041: Obtain spare parts inventory ordering data, which includes fixed cost of a single spare parts order, cost of a single spare parts order, current spare parts inventory, current spare parts in transit inventory, and spare parts forecast cycle. Step 1042: The target inventory within the spare parts forecast period is the minimum spare parts basic inventory that meets the spare parts service level. ; Step 1042: Calculate the optimal inventory order quantity within the spare parts forecast period and determine the minimum inventory order cost within the spare parts forecast period: in, is the optimal inventory order quantity within the spare parts forecast period, is the minimum inventory ordering cost within the spare parts forecast period, Fixed cost for a single order of spare parts, is the single ordering cost of spare parts, is the forecast period, is the target inventory within the spare parts forecast period, is the current inventory of spare parts; N is the current in-transit inventory of spare parts.

11. A device for predicting spare parts demand for nuclear power plants based on Bayesian reasoning, comprising a memory and a processor, wherein the memory stores computer-readable instructions, characterized in that: When the processor executes the computer-readable instructions, the processor implements the steps of the method for predicting spare parts demand for nuclear power plants based on Bayesian reasoning according to any one of claims 1 to 10.

Images