Personalized situation-based maintenance strategy optimization method and system for shared rental equipment
Through Monte Carlo simulation and grid search algorithm, combined with Poisson process and Gamma stochastic process, the maintenance strategy of shared rental equipment is optimized, and the maintenance cost problem caused by uncertain equipment utilization in shared rental services is solved, and the equipment is realized Effective balance between performance and cost.
Patent Information
- Application Number
- CN202411048501.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-01
- Publication Date
- 2024-11-08
- Estimated Expiration
- 2044-08-01
AI Technical Summary
In the shared rental service scenario, the uncertainty of equipment usage and unpredictable return time make it difficult to design and execute effective maintenance plans on the situation, increasing maintenance costs and management challenges.
Using Monte Carlo simulation and grid search algorithm, combined with Poisson process, Gamma stochastic process and AFT model, personalized state detection scheme and maintenance strategy are constructed, and equipment maintenance strategy is optimized through Monte Carlo simulation and grid search to reduce equipment life The average cost rate for the cycle.
It realizes that while meeting equipment performance requirements, it significantly reduces maintenance costs, improves the matching degree between maintenance activities and equipment status, enhances the scientificity and accuracy of decision-making, and optimizes the maintenance strategy of shared rental equipment.
Smart Images

Figure CN118569107B_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of equipment system condition-based maintenance strategy optimization, and specifically relates to a method and system for optimizing a personalized condition-based maintenance strategy for shared rental equipment based on Monte Carlo simulation and grid search. Background Art
[0002] In recent years, as the concept of sharing economy has gained popularity, many manufacturing companies have entered the sharing market, not only selling products, but also starting to provide diversified shared leasing services. In this service-oriented business model, manufacturers, as the owners of equipment, are responsible for the operation and maintenance costs of the equipment and provide the equipment to consumers on demand. Consumers obtain the right to use the equipment through flexible and diverse payment models (such as time-based leasing, time-sharing leasing, and pay-per-use), and return the equipment after completing the required tasks for other users to recycle. For example, Chinese new energy vehicle company NIO launched an innovative BaaS (Battery as a Service) business model in 2020, which is to lease power batteries as a shared resource. Customers can sign a battery leasing agreement and pay the corresponding leasing fee according to their own needs. With the help of NIO's replaceable battery technology and a network of battery swap stations across the country, car owners can easily replace exhausted batteries with fully charged batteries of the same capacity. At the same time, NIO is responsible for the maintenance of the battery throughout the process to ensure the good operation of each battery.
[0003] Compared with the traditional static long-term leasing model, this new type of shared leasing service presents a significant dynamic short-term sharing feature - a single device will be used continuously by multiple unspecified users (lessees) throughout its life cycle, and user replacement is frequent and random. In the face of this trend, how to design a set of personalized condition-based maintenance strategies that can not only ensure the stable and reliable operation of equipment but also effectively control maintenance costs, and help operators stand out in the fiercely competitive market environment, has become a key issue that needs to be solved urgently. With the development of advanced technologies such as the Internet of Things and big data, condition-based maintenance strategies are increasingly receiving high attention from academia and industry. Condition-based maintenance strategies monitor and evaluate the actual operating status of equipment in real time, and then formulate efficient and reasonable maintenance plans in a targeted manner, taking into account both economic benefits and maintenance effects. However, in the shared service scenario, equipment maintenance management is more challenging due to problems such as uncertain utilization rate and unpredictable return time.
[0004] Based on the above situation, the present invention combines Monte Carlo simulation and grid search to provide equipment operators with an effective technical solution to achieve personalized condition-based maintenance for the characteristics of shared rental equipment, thereby significantly reducing maintenance costs while meeting equipment performance requirements. Summary of the invention
[0005] In order to solve the difficulties in the maintenance of existing shared equipment, namely, it is difficult to design and implement effective condition-based maintenance plans due to fluctuations in usage frequency and uncertainty in return time, the present invention proposes an innovative personalized condition-based maintenance strategy optimization technical solution, which uses Monte Carlo simulation technology and grid search algorithm, and can adaptively formulate equipment maintenance plans based on the characteristics of shared rental equipment. While ensuring equipment performance requirements, it minimizes the average cost rate of the equipment life cycle, thereby achieving efficient resource management and cost control.
[0006] In order to achieve the above-mentioned object of the invention, the present invention adopts the following technical solutions:
[0007] The method for optimizing the personalized condition-based maintenance strategy of shared rental equipment includes the following steps:
[0008] S1, constructs the usage rate distribution function of the user group based on the historical sensor data of the shared rental equipment;
[0009] S2, uses the Poisson process to describe the behavior of customers actively changing equipment in the shared rental scenario;
[0010] S3, using the Gamma random process and AFT model to describe the characteristics of equipment degradation and determine the mechanism of the effect of maintenance strategy on the degradation process of shared rental equipment;
[0011] S4, establish a personalized status detection plan;
[0012] S5, establish the implementation methods of preventive maintenance and corrective maintenance strategies;
[0013] S6, based on the renewal theory, obtains the average cost rate model under the long-term operation conditions of the equipment, solves the model through Monte Carlo simulation and grid search, and obtains the optimization solution of the condition-based maintenance strategy.
[0014] Furthermore, step S1 specifically includes the following steps:
[0015] S11, collecting historical sensor data of shared rental equipment, including but not limited to usage frequency and usage pattern;
[0016] S12, select different usage rate distribution functions for fitting and perform parameter estimation;
[0017] S13, determining the optimal usage probability density function.
[0018] Further, step S2 includes the following steps:
[0019] S21, collect historical data of shared rental equipment, including frequency and time information of customers changing equipment;
[0020] S22, based on the historical usage time data, construct a Poisson process model, specifically:
[0021] The intensity parameter used is Homogeneous Poisson process Simulate the randomness of customers actively changing equipment in a shared rental scenario; use Indicates the time interval The total number of times the device has changed hands, for any time greater than 0 For example, is a non-negative random variable with parameters The Poisson distribution of , the probability distribution function is:
[0022] (1)
[0023] in, is an integer greater than or equal to 0;
[0024] The switching intensity of different customers With its utilization rate The relationship is:
[0025] (2)
[0026] Among them, the parameters of the Poisson process Indicated in nominal utilization rate The average number of times customers change equipment per unit time;
[0027] S23, performing parameter estimation on the Poisson process model in step S22.
[0028] Further, step S3 includes the following steps:
[0029] S31, the state of the device at time t is represented by a random variable By analyzing the relevant properties of the Gamma process and combining the collected equipment performance data, the Gamma process is used to represent the equipment degradation process. rationality;
[0030] Nominal utilization rate If there is no maintenance measure, for all , increment Obey the scale parameter and shape parameters The gamma distribution of >0, the shape parameters and scale parameters of the Gamma degradation process are determined using parameter estimation based on the performance data of the equipment degradation process;
[0031] S32, based on the equipment maintenance history and customer usage data, including maintenance time, maintenance type (preventive maintenance, corrective maintenance, etc.) and maintenance results, as well as equipment usage data and other related data, construct an AFT model to describe the relationship between equipment degradation characteristics and maintenance measures and equipment usage;
[0032] Maintenance measures include preventive maintenance measures and corrective maintenance measures; preventive maintenance measures include perfect preventive maintenance and imperfect preventive maintenance; perfect preventive maintenance restores the equipment to a brand new ideal state; imperfect preventive maintenance restores the equipment to a better state between good as new and bad as old; corrective maintenance is carried out after the equipment fails, used to deal with the equipment failure and restore the equipment performance to a brand new state; in the case of the above maintenance measures, the degradation characteristics of the equipment will be affected, as follows:
[0033] Various maintenance measures have an impact on the degradation rate of equipment. As the number of imperfect preventive maintenance on the equipment increases, the equipment degradation rate accelerates. The device is in The average degradation rate of the equipment after the imperfect preventive maintenance action, where represents the number of consecutive imperfect preventive maintenance, , the effect of imperfect preventive maintenance on equipment degradation rate is represented by a non-negative continuous random variable To describe, the corresponding density probability distribution is , No. After imperfect preventive maintenance actions, the average degradation rate of the equipment becomes , No. After the imperfect preventive maintenance action, the shape parameter of the equipment degradation process is , if the maintenance action is perfect preventive maintenance or corrective maintenance, the average degradation rate of the equipment is reset to ;
[0034] Various maintenance measures have an impact on the degradation level of the equipment. After imperfect preventive maintenance, the equipment recovers from the degradation state before maintenance to a non-zero random state. The impact of imperfect preventive maintenance on equipment performance is set with a random improvement factor. To characterize; if at the time of inspection Detect device status , then execute An imperfect preventive maintenance action. , the critical performance level of the equipment becomes , different improvement factors are independent and identically distributed, and their probability density function is set as , and the improvement factor , The probability density function of is determined by collecting the maintenance records of the equipment, selecting different functions for fitting, and using parameter estimation; after perfect preventive maintenance or corrective maintenance, the equipment degradation level becomes 0, that is, it is restored to a brand new state;
[0035] The utilization rate will affect the degradation rate of the device. When the device is running at a high utilization rate, the degradation rate will be accelerated. The utilization rate is considered when simulating device degradation, as follows:
[0036] set up Consider continuous The shape parameters of the equipment degradation process obtained by considering the influence of sub-imperfect preventive maintenance behavior and usage rate on equipment degradation are: ,in Indicates the AFT factor, which is the utilization rate and using the acceleration factor function, for all , increment Obey the scale parameter and shape parameters Gamma distribution:
[0037] (3)
[0038] in, ; is an indicator function, if ,but , otherwise 0;
[0039] S33, performing parameter estimation on the Gamma process in step S31 and the AFT model in step S32;
[0040] S34, estimating the probability density function of the random variables involved in the impact of the imperfect preventive maintenance on the equipment degradation process described in step S32 by using the maximum likelihood estimation method.
[0041] Further, step S4 includes the following steps:
[0042] S41, determining the equipment remaining useful life distribution function according to the user information, setting the equipment remaining useful life distribution function The value corresponding to the quantile is the next planned detection interval. is the performance risk threshold, which is a quantity to be decided;
[0043] The remaining service life RUL is defined as the state of the equipment at the current time t. , until the state degrades to below the performance guarantee threshold Remaining working time:
[0044] (4)
[0045] set up Indicates the device At the time of inspection, the corresponding equipment degradation level is ; Indicates The cumulative number of consecutive imperfect preventive maintenance after inspection and maintenance decision execution; the next inspection time is expressed as:
[0046] (5)
[0047] in, , is the cumulative distribution function of RUL, It is RUL distribution Quantile:
[0048] (6)
[0049] is the incomplete Gamma function, , , ;
[0050] For a linear equation ; Use the bisection method to solve the unknown The value of is the next planned inspection interval;
[0051] S42, dynamically adjust the detection time according to the time when the user randomly changes the device. If the user returns the device before the next scheduled detection time or returns the device at the time of the next detection, the detection will be carried out in advance or at the time, and the next detection time will be set according to the usage rate of the next user; if the user does not return the device before the next detection, the detection will not be carried out in advance, and the time of the next detection will still be the originally planned detection time. After the detection at the originally planned detection time, the next detection time will continue to be set based on the current user situation.
[0052] Further, step S5 includes the following steps:
[0053] S51, determine the performance guarantee threshold of the device When the degradation level of the equipment exceeds the pre-specified performance guarantee threshold When the device is in a state of low power consumption, the performance of the equipment will be seriously damaged;
[0054] S52: It is detected that the degradation level of the equipment is greater than or equal to the preventive maintenance threshold but less than the performance guarantee threshold. , preventive maintenance is required, It is a quantity to be decided. Preventive maintenance is preferably imperfect preventive maintenance. As the number of imperfect preventive maintenance increases, the degradation rate of the equipment will gradually accelerate. It is required to be carried out continuously. After the first imperfect preventive maintenance, Perfect preventive maintenance is only used when the first preventive maintenance is required; is the imperfect maintenance number threshold, which is a quantity to be decided;
[0055] S53: When it is detected that the degradation level of the device is greater than or equal to the pre-specified performance guarantee threshold, , equipment performance loss is serious and corrective maintenance is carried out.
[0056] Further, step S6 includes the following steps:
[0057] S61, the cost incurred by the equipment due to maintenance activities during a life cycle is obtained through the expected value of the random variable to obtain the final average cost rate model;
[0058] The costs involved in calculating the average cost rate of equipment under operation within a set period include total inspection costs, total preventive maintenance costs, total corrective maintenance costs and total performance breach costs. , , They are the cost of a single inspection, the cost of a single corrective maintenance, and the cost of a single perfect preventive maintenance. The penalty fee per unit time for continuing to run after exceeding the performance guarantee threshold;
[0059] The cost of imperfect preventive maintenance is modeled as a function of the improvement factor, and the cost of each IPM is defined as:
[0060] (7)
[0061] No. The improvement factor of an IPM is , is a natural number, is a non-negative real number, It is the IPM cost when the performance degradation level of the equipment drops to 0 under imperfect preventive maintenance measures;
[0062] S62, assume that the operation time of the equipment is infinite. According to the basic renewal theory, the average cost rate of the equipment in infinite operation time is equal to the average cost rate of the equipment in one life cycle. , , , , They are If the total number of IPMs, PPMs, CMs, inspections, and total time of continuous operation exceeds the performance guarantee threshold, the device will run until The total maintenance cost at this moment is:
[0063] (8)
[0064] Using the long-term expected maintenance cost rate as a criterion, the optimal decision variable is finally determined by minimizing the long-term average operating cost rate. , , The value of , using the renewal theory, the long-term expected maintenance cost per unit time is:
[0065] (9)
[0066] Indicates the length of the equipment's life cycle. The time it takes to replace the equipment twice to a brand new state is defined as one life cycle. represents the mathematical expectation of the equipment life cycle, is the mathematical expectation of the total maintenance cost of the equipment during its life cycle;
[0067] S63, setting the number of simulations for Monte Carlo simulation , Monte Carlo simulation is used to approximately solve the corresponding average cost rate for each set of decision variables;
[0068] S64, based on the grid search method, in the set parameter range by step size , , Search for the optimal maintenance strategy and obtain the optimal solution for the condition-based maintenance strategy, including the preventive maintenance threshold, the imperfect maintenance number threshold, and the performance risk threshold.
[0069] The present invention also discloses a system for optimizing the personalized maintenance strategy of shared rental equipment based on the above method, which includes the following modules:
[0070] Usage distribution function building module: builds the usage distribution function of the user group based on the historical sensor data of the shared rental equipment;
[0071] Poisson process description module: Use Poisson process to describe the behavior of customers actively changing equipment in shared rental scenarios;
[0072] Equipment degradation feature description module: Use Gamma random process and AFT model to describe the characteristics of equipment degradation and determine the mechanism of the effect of maintenance strategy on the degradation process of shared rental equipment;
[0073] Detection plan establishment module: establish personalized status detection plan;
[0074] Maintenance strategy establishment module: Establish the implementation methods of preventive maintenance and corrective maintenance strategies;
[0075] Condition-based maintenance strategy optimization solution acquisition module: Based on the renewal theory, the average cost rate model of shared rental equipment under long-term operating conditions is obtained, and the condition-based maintenance strategy optimization solution is obtained through Monte Carlo simulation and grid search to solve the model.
[0076] In summary, the present invention proposes a method and system for optimizing the personalized situation-based maintenance strategy of shared rental equipment based on Monte Carlo simulation and grid search algorithm. The Monte Carlo method, also known as random sampling or statistical experimental method, originated in the 1940s and developed as part of computational mathematics. In many real-world problems, traditional analytical methods often fail to provide accurate answers, which led to the birth of the Monte Carlo method. It is a numerical calculation strategy based on probability and statistical theory, and its core lies in using random number generation technology to simulate and understand the behavior of complex systems. Monte Carlo simulation is known for its experimental characteristics and relies on the principle of the law of large numbers to solve mathematical problems through random numbers. When it is necessary to estimate the probability of an event or calculate the expected value of a random variable, the Monte Carlo method obtains probability estimates or expected values by constructing random experiments, thereby providing an approximate solution to the problem. The key steps of Monte Carlo simulation include: 1) constructing and describing the probability process; 2) performing random sampling on a given probability distribution; 3) calculating the required estimated values.
[0077] The grid search method is an optimization algorithm often used in machine learning. It is an exhaustive search algorithm that traverses the specified parameter model. The meaning of grid search is to discretize the continuous variables within the specified range, divide them into a finite number of variable values according to the set grid density (or step size), and then substitute the current values of the variables into the model to calculate the objective function value at this time. In this way, the grid search method can find the parameter combination that makes the objective function reach the optimal value within the set parameter range.
[0078] The present invention provides a more refined situation-based maintenance strategy to address the problems of random usage rates and unpredictable return times of shared rental equipment, overcoming the difficulties in maintenance decision-making caused by such uncertain factors in the prior art. The present invention realizes personalized customization of equipment maintenance strategies, and can dynamically adjust the timing and methods of maintenance according to the actual usage of different equipment, thereby improving the matching degree between maintenance activities and the actual status of the equipment; Monte Carlo simulation technology is used to simulate a variety of possible equipment usage scenarios and failure modes, providing comprehensive data support for the formulation of maintenance strategies, and enhancing the scientificity and accuracy of decision-making; combined with the grid search algorithm, the present invention can find the optimal maintenance strategy that minimizes the average maintenance cost rate over the entire life cycle of the equipment while meeting the equipment performance indicators. BRIEF DESCRIPTION OF THE DRAWINGS
[0079] Figure 1 A flowchart of a method for optimizing a personalized condition-based maintenance strategy for shared rental equipment based on Monte Carlo simulation and grid search according to a preferred embodiment of the present invention;
[0080] Figure 2 A schematic diagram of the gamma degradation process of the equipment in the present invention without maintenance measures;
[0081] Figure 3 A schematic diagram of the change in equipment degradation rate due to various maintenance measures in the present invention;
[0082] Figure 4 A schematic diagram of the implementation method of the preventive maintenance and corrective maintenance strategies in the present invention and the mechanism of action of these strategies on the equipment degradation level;
[0083] Figure 5 A schematic diagram of formulating the next planned detection interval in the present invention;
[0084] Figure 6 A schematic diagram of a solution for dynamically adjusting personalized status detection in the present invention;
[0085] Figure 7 A schematic diagram of the Monte Carlo simulation in the present invention;
[0086] Figure 8 This is a block diagram of a system for optimizing the personalized situation-based maintenance strategy for shared rental equipment based on Monte Carlo simulation and grid search in a preferred embodiment of the present invention. DETAILED DESCRIPTION
[0087] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.
[0088] The present invention is further illustrated below in conjunction with specific embodiments. Those skilled in the art should understand that these embodiments are only used to illustrate the present invention and are not used to limit the scope of the present invention, and modifications to various equivalent forms of the present invention fall within the scope defined by the appended claims of this application.
[0089] Embodiment 1:
[0090] like Figure 1 As shown, this embodiment provides a method for optimizing a personalized condition-based maintenance strategy for shared rental equipment based on Monte Carlo simulation and grid search, including the following steps:
[0091] 1. Based on the historical sensor data of shared rental equipment, construct the usage rate distribution function of the user group;
[0092] 2. Use Poisson process to describe the behavior of customers actively changing equipment in shared rental scenarios;
[0093] 3. Use the Gamma random process and AFT model to describe the characteristics of equipment degradation and determine the mechanism of the effect of maintenance strategy on the degradation process of shared rental equipment;
[0094] 4. Establish a personalized status detection plan;
[0095] 5. Establish the implementation methods of preventive maintenance and corrective maintenance strategies;
[0096] 6. Based on the renewal theory, the average cost rate model of shared rental equipment under long-term operation conditions is obtained. Through Monte Carlo simulation and grid search to solve the model, the optimization solution of the condition-based maintenance strategy is obtained.
[0097] The following is a detailed description of each step.
[0098] Step 1 specifically includes the following steps:
[0099] 1-1, collect historical sensor data of shared rental equipment, including but not limited to usage frequency and usage patterns;
[0100] 1-2, select different usage rate distribution functions for fitting and perform parameter estimation;
[0101] 1-3, determine the optimal utilization probability density function, assuming that the device is in the The device usage rate of each customer is , is a random variable in the interval Contains The probability density function of .
[0102] Step 2 specifically includes the following steps:
[0103] 2-1, collect historical data of shared rental equipment, including the frequency and time information of customers changing equipment;
[0104] 2-2, based on the historical usage time data, construct a Poisson process model, specifically:
[0105] The intensity parameter used is Homogeneous Poisson process Simulate the randomness of customers actively changing equipment in a shared rental scenario; use Indicates the time interval The total number of times the device has changed hands, for any time greater than 0 For example, is a non-negative random variable with parameters The Poisson distribution of , the probability distribution function is:
[0106] (1)
[0107] in, is an integer greater than or equal to 0;
[0108] Due to different customer usage rates, the intensity of replacing equipment is also different. The replacement intensity of different customers With its utilization rate The relationship is:
[0109] (2)
[0110] Among them, the parameters of the Poisson process Indicated in nominal utilization rate The average number of times customers change equipment per unit time;
[0111] 2-3. Estimate the parameters of the Poisson process model in step 2-2.
[0112] Step 3 specifically includes the following steps:
[0113] 3-1, the state (degradation level) of the equipment at time t is represented by a random variable By analyzing the relevant properties of the Gamma process and combining the collected equipment performance data, such as performance degradation and maintenance records, the Gamma process is used to represent the equipment degradation process. rationality;
[0114] like Figure 2 As shown, assuming the nominal utilization rate If there is no maintenance measure, for all , increment Obey the scale parameter and shape parameters The gamma distribution of >0, the shape parameters and scale parameters of the Gamma degradation process are determined using parameter estimation based on the performance data of the equipment degradation process;
[0115] 3-2, based on the equipment maintenance history and customer usage data, including maintenance time, maintenance type (preventive maintenance, corrective maintenance, etc.) and maintenance results, as well as equipment usage data and other related data, construct an AFT model to describe the relationship between equipment degradation characteristics and maintenance measures and equipment usage;
[0116] Maintenance measures mainly include preventive maintenance measures and corrective maintenance measures. Preventive maintenance is a kind of pre-emptive maintenance, which aims to ensure the operating performance of the equipment to reduce the risk of failure. According to the different maintenance effects, preventive maintenance measures can be divided into perfect preventive maintenance and imperfect preventive maintenance. Perfect preventive maintenance restores the equipment to a brand new ideal state, while imperfect preventive maintenance restores the equipment to a better state between good as new and bad as old. Corrective maintenance is carried out after the equipment fails. Its purpose is to deal with the equipment failure and restore its performance to a brand new state. Therefore, with these maintenance measures, the degradation characteristics of the equipment will be affected, as follows:
[0117] Various maintenance measures have an impact on the degradation rate of equipment, such as Figure 3 As shown in Figure 2, as the number of imperfect preventive maintenance on the equipment increases, the equipment degradation rate accelerates. The device is in The average degradation rate of the equipment after the imperfect preventive maintenance action, where represents the number of consecutive imperfect preventive maintenance, , the effect of imperfect preventive maintenance on equipment degradation rate is represented by a non-negative continuous random variable To describe, the corresponding density probability distribution is , No. After imperfect preventive maintenance actions, the average degradation rate of the equipment becomes , No. After the imperfect preventive maintenance action, the shape parameter of the equipment degradation process is , if the maintenance action is perfect preventive maintenance or corrective maintenance, the average degradation rate of the equipment is reset to ;
[0118] Various maintenance measures have an impact on the degradation level of equipment. The specific mechanisms of the impact of preventive maintenance and corrective maintenance strategies on the degradation level of equipment are as follows: Figure 4 As shown in the figure, after imperfect preventive maintenance, the equipment recovers from the degraded state before maintenance to a non-zero random state. The impact of imperfect preventive maintenance on equipment performance is set with a random improvement factor To characterize; if at the time of inspection Detect device status , then execute An imperfect preventive maintenance action. , the critical performance level of the equipment becomes , different improvement factors are independent and identically distributed, and their probability density function is set as , and the improvement factor , The probability density function of is determined by collecting the maintenance records of the equipment, selecting different functions for fitting, and using parameter estimation; after perfect preventive maintenance or corrective maintenance, the equipment degradation level becomes 0, that is, it is restored to a brand new state;
[0119] The utilization rate will affect the degradation rate of the device. When the device is running at a high utilization rate, the degradation rate will be accelerated. The utilization rate is considered when simulating device degradation, as follows:
[0120] set up Consider continuous The shape parameters of the equipment degradation process obtained by considering the influence of sub-imperfect preventive maintenance behavior and usage rate on equipment degradation are: ,in Indicates the AFT factor, which is the utilization rate and using the acceleration factor function, for all , increment Obey the scale parameter and shape parameters Gamma distribution:
[0121] (3)
[0122] in, ; is an indicator function, if ,but , otherwise 0;
[0123] 3-3, estimate the parameters of the equipment Gamma degradation model in step 3-1 and the AFT model in step 3-2;
[0124] 3-4. Regarding the impact of imperfect preventive maintenance on the equipment degradation process described in step 3-2, the probability density function of the random variables involved is estimated by the maximum likelihood estimation method.
[0125] Step 4 specifically includes the following steps:
[0126] 4-1, determine the equipment remaining service life distribution function based on user information, and set the equipment remaining service life distribution function. The value corresponding to the quantile is the next planned detection interval. is the performance risk threshold, which is a quantity to be decided; the method for formulating the next planned detection interval is as follows Figure 5 As shown;
[0127] The remaining service life RUL is defined as the state of the equipment at the current time t. , until the state degrades to below the performance guarantee threshold Remaining working time:
[0128] (4)
[0129] set up Indicates the device At the time of inspection, the corresponding equipment degradation level is (If in If maintenance actions are performed at all times, the degradation level of the equipment after maintenance); Indicates The cumulative number of consecutive imperfect preventive maintenance after the inspection and maintenance decision (historical PM records, including the The next inspection time is expressed as:
[0130] (5)
[0131] in, , is the cumulative distribution function of RUL, It is RUL distribution Quantile:
[0132] (6)
[0133] is the incomplete Gamma function, , , ;
[0134] For a linear equation ; Use the bisection method to solve the unknown The value of is the interval between the next planned detection; Binary search is also called binary search, which is a search algorithm for finding a specific element in an ordered array. The basic idea of binary search is to divide the array into two halves. By comparing the size of the middle element with the target value, half of the elements that do not contain the target value can be excluded, thereby reducing the search range by half, and repeating this process until the target value is found or the range is reduced to 0;
[0135] 4-2, dynamically adjust the detection time according to the time when the user randomly changes the device. If the user returns the device before the next detection time originally planned or returns the device at the time of the next detection, it is necessary to conduct the detection in advance or at the time, and then set the next detection time according to the usage rate of the next user; if the user does not return the device before the next detection, it is not necessary to conduct the detection in advance, and the time of the next detection is still the original detection time. After the detection at the original planned detection time, the next detection time will continue to be set based on the current user situation; Dynamic adjustment of personalized status detection schemes such as Figure 6 shown.
[0136] Step 5 specifically includes the following:
[0137] 5-1. Determine the performance guarantee threshold of the equipment When the degradation level of the equipment exceeds the pre-specified performance guarantee threshold When the device is faulty, the performance of the device is seriously impaired. This soft fault will not sound automatically and can only be discovered through detection. The period from when the performance deficiency occurs to when it is discovered is the time the device performance deficiency has passed, which will incur additional penalty costs for the operator.
[0138] 5-2, the equipment degradation level is detected to be greater than or equal to the preventive maintenance threshold but less than the performance guarantee threshold. , preventive maintenance is required, It is a quantity to be decided. Preventive maintenance is preferably imperfect preventive maintenance. As the number of imperfect preventive maintenance increases, the degradation rate of the equipment will gradually accelerate. It is required to be carried out continuously. After the first imperfect preventive maintenance, Perfect preventive maintenance is only used when the first preventive maintenance is required; is the imperfect maintenance number threshold, which is a quantity to be decided;
[0139] 5-3, when it is detected that the degradation level of the equipment is greater than or equal to the pre-specified performance guarantee threshold , the equipment performance loss is serious and perfect corrective maintenance (CM) must be carried out.
[0140] Step 6 specifically includes the following steps:
[0141] 6-1, the cost incurred by maintenance activities during the life cycle of the equipment is obtained through the expected value of the random variable to obtain the final average cost rate model;
[0142] The costs involved in calculating the average cost rate of equipment under operation within a set period include total inspection costs, total preventive maintenance costs, total corrective maintenance costs and total performance breach costs. , , They are the cost of a single inspection, the cost of a single corrective maintenance, and the cost of a single perfect preventive maintenance. The penalty fee per unit time for continuing to run after exceeding the performance guarantee threshold;
[0143] Imperfect preventive maintenance costs can be modeled as a function of the improvement factor. Essentially, the improvement factor The higher it is, the greater the maintenance workload and the higher the maintenance cost. Then the cost of each IPM is defined as:
[0144] (7)
[0145] No. The improvement factor of an IPM is , is a natural number, is a non-negative real number, It is the IPM cost when the performance degradation level of the equipment drops to 0 under imperfect preventive maintenance measures. This cost is usually lower than the perfect preventive maintenance cost.
[0146] 6-2, assume that the operation time of the equipment is infinite. According to the basic renewal theory, the average cost rate of the equipment in infinite operation time is equal to the average cost rate of the equipment in one life cycle. , , , , They are If the total number of IPMs, PPMs, CMs, inspections, and total time of continuous operation exceeds the performance guarantee threshold, the device will run until The total maintenance cost at this moment is:
[0147] (8)
[0148] In order to evaluate the performance of the maintenance strategy, the long-term expected maintenance cost rate is used as the criterion. The goal is to minimize the long-term average cost rate to ultimately determine the optimal decision variable. , , The value of , using the renewal theory, the long-term expected maintenance cost per unit time is:
[0149] (9)
[0150] Indicates the length of the equipment's life cycle. The time it takes to replace the equipment twice to a brand new state is defined as one life cycle. represents the mathematical expectation of the equipment life cycle, is the mathematical expectation of the total maintenance cost of the equipment during its life cycle;
[0151] The constraints of the model are:
[0152] Constraint 1: The preventive maintenance threshold has upper and lower limit constraints: ;
[0153] Constraint 2: The threshold of the number of imperfect preventive maintenance has upper and lower limit constraints: ;
[0154] Constraint 2: The performance guarantee threshold has upper and lower limit constraints: ;
[0155] 6-3, Set the number of simulations for Monte Carlo simulation , the Monte Carlo simulation method can be used to approximately solve the corresponding average cost rate for each set of decision variables. The specific Monte Carlo simulation process is as follows: Figure 7 As shown;
[0156] 6-4, based on the grid search method, the step size is set within the set parameter range. , , Search for the optimal maintenance strategy and obtain the optimal solution for the condition-based maintenance strategy, including the preventive maintenance threshold, the imperfect maintenance number threshold, and the performance risk threshold.
[0157] The following is a specific implementation case of optimizing the shared rental equipment maintenance strategy using the optimization method of the present invention. Consider a shared equipment whose performance will degrade over time; the usage rate of the shared equipment customer group Gamma distribution , shape parameter , scale parameter , the value range is 0.5 to 6.50 ; The nominal utilization rate is set to ; About the average number of times customers replace equipment per unit time under nominal utilization rate ; The initial degradation level of the equipment is zero, that is , when no maintenance is performed, the relevant parameters of the Gamma degradation process are set to , , ; The impact of imperfect preventive maintenance on equipment degradation rate is expressed as a non-negative continuous random variable To describe, it obeys exponential distribution, and the density probability distribution is , is a non-negative real number, The mean value of ; About the parameters related to the AFT model ; The degradation level of the equipment exceeds the performance guarantee threshold , the equipment performance will be seriously damaged and the working capacity of the equipment will be reduced; the imperfect maintenance random improvement factor It follows a truncated normal distribution with a mean of The standard deviation is , About maintenance costs =1000 yuan / time, =9000 yuan / time, =7000 yuan / time, =12000 yuan / time, =2000 yuan / time, ; About grid search step size setting =0.1, =1, =0.01.
[0158] Find the optimal solution through grid search , , , the corresponding minimum long-term average cost rate per unit time ; In order to prove the superiority of the personalized condition-based maintenance strategy, it is compared with the condition-based maintenance strategy with fixed detection interval and the condition-based maintenance strategy under average utilization rate, and the relevant results are shown in Table 1; When other parameters are set the same, the maintenance cost under the personalized condition-based maintenance strategy is saved by 22.17% and 13.04% respectively; therefore, the personalized condition-based maintenance strategy will bring better results.
[0159] Table 1 The optimal condition-based maintenance strategy and the corresponding minimum average cost rate under three strategies
[0160]
[0161] Embodiment 2:
[0162] like Figure 8 As shown, this embodiment discloses a system for optimizing a personalized maintenance strategy for shared rental equipment, based on the method of Embodiment 1, including the following modules:
[0163] Usage distribution function building module: builds the usage distribution function of the user group based on the historical sensor data of the shared rental equipment;
[0164] Poisson process description module: Use Poisson process to describe the behavior of customers actively changing equipment in shared rental scenarios;
[0165] Equipment degradation feature description module: Use Gamma random process and AFT model to describe the characteristics of equipment degradation and determine the mechanism of the effect of maintenance strategy on the degradation process of shared rental equipment;
[0166] Detection plan establishment module: establish personalized status detection plan;
[0167] Maintenance strategy establishment module: Establish the implementation methods of preventive maintenance and corrective maintenance strategies;
[0168] Condition-based maintenance strategy optimization solution acquisition module: Based on the renewal theory, the average cost rate model of shared rental equipment under long-term operating conditions is obtained, and the condition-based maintenance strategy optimization solution is obtained through Monte Carlo simulation and grid search to solve the model.
[0169] For other contents of this embodiment, please refer to Embodiment 1.
[0170] In summary, the present invention belongs to the technical field of condition-based maintenance of equipment systems, and specifically relates to a method and system for optimizing a personalized condition-based maintenance strategy for shared rental equipment based on Monte Carlo simulation and grid search, and the method is as follows: S1, constructing a usage distribution function of a user group based on historical sensor data of shared rental equipment; S2, using a Poisson process to describe the behavior of customers actively replacing equipment in a shared rental scenario; S3, using a Gamma random process and an AFT model to describe the characteristics of equipment degradation, and determining the mechanism of action of maintenance strategies on the degradation process of shared rental equipment; S4, establishing a personalized state detection plan; S5, establishing implementation methods for preventive maintenance and corrective maintenance strategies; S6, obtaining an average cost rate model under long-term operating conditions of shared rental equipment based on the renewal theory, and solving the model through Monte Carlo simulation and grid search to obtain a condition-based maintenance strategy optimization plan. The present invention can effectively realize personalized condition-based maintenance based on the characteristics of shared rental equipment, thereby significantly reducing maintenance costs while meeting equipment performance requirements.
[0171] The above is only a specific implementation of the present invention, but the protection scope of the present invention is not limited thereto. Any modification, equivalent replacement and improvement made by any person skilled in the art within the technical scope disclosed by the present invention and within the spirit and principle of the present invention shall be covered within the protection scope of the present invention. At the same time, the contents not described in detail in this specification belong to the prior art known to those skilled in the art.
Claims
1. Personalized maintenance strategy optimization method for shared rental equipment, It is characterized in that The steps include: S1, constructs the usage rate distribution function of the user group based on the historical sensor data of the shared rental equipment; S2, uses the Poisson process to describe the behavior of customers actively changing equipment in the shared rental scenario; S3, using the Gamma random process and AFT model to describe the characteristics of equipment degradation and determine the mechanism of the effect of maintenance strategy on the degradation process of shared rental equipment; S4, establish a personalized status detection plan; S5, establish the implementation methods of preventive maintenance and corrective maintenance strategies; S6, based on the renewal theory, obtain the average cost rate model of shared rental equipment under long-term operation conditions, solve the model through Monte Carlo simulation and grid search, and obtain the optimization solution of condition-based maintenance strategy; Step S2 includes the following steps: S21, collect historical data of shared rental equipment, including frequency and time information of customers changing equipment; S22, based on the historical usage time data, construct a Poisson process model, specifically: Use the strength parameter λ j The homogeneous Poisson process {M(t), t≥0} simulates the randomness of customers actively changing devices in the shared rental scenario; M(t) represents the total number of device changes in the time interval (0, t]. For any time t greater than 0, M(t) is a non-negative random variable with parameter λ j The Poisson distribution of t has a probability distribution function of: Wherein, m is an integer greater than or equal to 0; The switching intensity of different customers λ j With its utilization rate U j The relationship is: Among them, the parameter λ of the Poisson process 0 Indicates the nominal utilization rate U 0 The average number of times customers change equipment per unit time; S23, performing parameter estimation on the Poisson process model in step S22; Step S3 includes the following steps: S31, the state of the device at time t is represented by a random variable X t By analyzing the relevant properties of the Gamma process and combining the collected equipment performance data, the equipment degradation process X is represented by the Gamma process. t the rationality of Assuming the nominal utilization rate U 0 If there is no maintenance measure, for all 0≤h≤t, the increment X t -X h Subject to the scale parameter β and shape parameter α 0 The gamma distribution of (th), β>0, the shape parameter and scale parameter of the gamma degradation process are determined using parameter estimation based on the performance data of the equipment degradation process; S32, based on the equipment maintenance history and customer usage data, construct an AFT model to describe the relationship between equipment degradation characteristics and maintenance measures and equipment usage rate; Maintenance measures include preventive maintenance measures and corrective maintenance measures; preventive maintenance measures include perfect preventive maintenance and imperfect preventive maintenance; perfect preventive maintenance restores the equipment to a brand new ideal state; imperfect preventive maintenance restores the equipment to a better state between good as new and bad as old; corrective maintenance is carried out after the equipment fails, used to deal with the equipment failure and restore the equipment performance to a brand new state; in the case of the above maintenance measures, the degradation characteristics of the equipment will be affected, as follows: Various maintenance measures have an impact on the degradation rate of equipment. As the number of imperfect preventive maintenance on the equipment increases, the degradation rate of the equipment accelerates. k is the average degradation rate of the equipment after the kth imperfect preventive maintenance action, where k represents the number of consecutive imperfect preventive maintenance actions, k = 1, 2, ..., N, and the impact of imperfect preventive maintenance on the equipment degradation rate is represented by a non-negative continuous random variable ò k To describe, the corresponding density probability distribution is h(ò k ), after the kth imperfect preventive maintenance action, the average degradation rate of the equipment becomes ν k =v k-1 +ò k , after the kth imperfect preventive maintenance action, the shape parameter of the equipment degradation process is α k =v k β, if the maintenance action is perfect preventive maintenance or corrective maintenance, the average degradation rate of the equipment is reset to ν 0 =α 0 / β; Various maintenance measures have an impact on the degradation level of the equipment. After imperfect preventive maintenance, the equipment recovers from the degradation state before maintenance to a non-zero random state. The impact of imperfect preventive maintenance on equipment performance is characterized by a random improvement factor ρ. If at the inspection time T i Detect device status is X Ti , then the kth imperfect preventive maintenance action is performed, k = 1, 2, ..., N, and the critical performance level of the equipment becomes X k =(1-ρ k )X Ti , different improvement factors are independent and identically distributed, and their probability density function is set to p(ρ), and the improvement factor ρ∈(0,1], the probability density function of ρ is determined by collecting the maintenance records of the equipment, selecting different functions for fitting, and using parameter estimation; after perfect preventive maintenance or corrective maintenance, the equipment degradation level becomes 0, that is, it is restored to a brand new state; S33, performing parameter estimation on the Gamma process in step S31 and the AFT model in step S32; S34, estimating the probability density function of the random variables involved in the impact of the imperfect preventive maintenance on the equipment degradation process described in step S32 by using the maximum likelihood estimation method; In step S32, the device usage rate affects the degradation speed of the device. When the device is running at a high usage rate, the degradation speed will be accelerated. The usage rate is considered when simulating device degradation, as follows: set up is the shape parameter of the equipment degradation process after considering the impact of k consecutive imperfect preventive maintenance behaviors and usage rate on equipment degradation, then in Indicates the AFT factor, which is the utilization rate U j and a function using an acceleration factor θ (θ ≥ 1), for all 0 ≤ h ≤ t, increment X t -X h Subject to the scale parameter β and shape parameter Gamma distribution: in, I {Δx≥0} is the indicator function. If Δx ≥ 0, then I {Δx≥0} =1, otherwise 0; Step S4 includes the following steps: S41, determining the equipment remaining useful life distribution function according to the user information, setting the value corresponding to the Q quantile of the equipment remaining useful life distribution as the next planned inspection interval, where Q is the performance risk threshold, which is a quantity to be decided; S42, dynamically adjusting the detection time according to the time when the user randomly changes the device. If the user returns the device before the originally planned next detection time or returns the device at the time of the next detection, the detection is carried out in advance or at the time, and then the next detection time is set according to the usage rate of the next user; if the user does not return the device before the next detection, the detection is not carried out in advance at this time, and the time of the next detection is still the originally planned detection time. After the detection according to the originally planned detection time, the next detection time is continued to be set based on the current user situation; Step S41 also includes the following steps: The remaining service life RUL is defined as the state of the equipment at the current time t is X t , until the state degrades to below the performance guarantee threshold l p Remaining working time: Assume T i Indicates the time when the equipment is inspected for the i-th time, and the corresponding equipment degradation level is n i It represents the cumulative number of consecutive imperfect preventive maintenance after the i-th inspection and maintenance decision execution; the next inspection time is expressed as: in, is the cumulative distribution function of RUL, is the Q quantile of the RUL distribution: Γ(s,w) is the incomplete Gamma function, For a linear equation Use the bisection method to solve the value of the unknown number ΔT, where ΔT is the next planned inspection interval; Step S5 includes the following steps: S51, determining the performance guarantee threshold l of the device p , when the degradation level of the equipment exceeds the pre-specified performance guarantee threshold l p When the device is in a state of low power consumption, the performance of the equipment will be seriously damaged; S52: It is detected that the degradation level of the equipment is greater than or equal to the preventive maintenance threshold but less than the performance guarantee threshold, i.e., pm ≤X t <l p , preventive maintenance is required, pm It is a quantity to be decided. Preventive maintenance prefers imperfect preventive maintenance. As the number of imperfect preventive maintenance increases, the degradation rate of the equipment will gradually accelerate. It is stipulated that after N consecutive imperfect preventive maintenance, perfect preventive maintenance will be adopted when the N+1th preventive maintenance is performed. N is the threshold of imperfect maintenance times, which is a quantity to be decided. S53, when it is detected that the degradation level of the device is greater than or equal to the pre-specified performance guarantee threshold, i.e., p ≤X t , equipment performance loss is serious, and corrective maintenance is performed; Step S6 includes the following steps: S61, the cost incurred by the equipment due to maintenance activities during a life cycle is obtained through the expected value of the random variable to obtain the final average cost rate model; The costs involved in calculating the average cost rate of equipment under the set period of operation include total inspection cost, total preventive maintenance cost, total corrective maintenance cost and total performance default cost, C i , C cm , C ppm They are the cost of a single inspection, the cost of a single corrective maintenance, and the cost of a single perfect preventive maintenance, C d The penalty fee per unit time for continuing to run after exceeding the performance guarantee threshold; The cost of imperfect preventive maintenance is modeled as a function of the improvement factor, and the cost of each IPM is defined as: The improvement factor of the kth IPM is ρ k , k is a natural number, δ is a non-negative real number, C ipm,0 It is the IPM cost when the performance degradation level of the equipment drops to 0 under imperfect preventive maintenance measures; S62, assume that the operation time of the equipment is infinite. According to the basic renewal theory, the average cost rate of the equipment in infinite operation time is equal to the average cost rate of the equipment in one life cycle. Let N ipm (t), N ppm (t), N cm (t), N i Where d(t) and d(t) are the total number of IPMs, PPMs, CMs, inspections, and the total time of continued operation beyond the performance guarantee threshold in [0, t], respectively. The total maintenance cost of the equipment running to time t is: Using the long-term expected maintenance cost rate as the standard, the optimal decision variable l is finally determined by minimizing the long-term average operating cost rate. pm , N, Q values, and using the update theory, the long-term expected maintenance cost per unit time is: L represents the life cycle of the equipment. The time from two consecutive replacements of the equipment to a brand new state is defined as a life cycle. E[L] represents the mathematical expectation of the equipment life cycle. E[C L (l pm ,N,Q)] is the mathematical expectation of the total maintenance cost of the equipment during its life cycle; S63, setting the number of simulations D for Monte Carlo simulation max , Monte Carlo simulation is used to approximately solve the corresponding average cost rate for each set of decision variables; S64, based on the grid search method, within the set parameter range, the step size is l pm_step 、N step , Q step Search for the optimal maintenance strategy and obtain the optimal solution for the condition-based maintenance strategy, including the preventive maintenance threshold, the imperfect maintenance number threshold, and the performance risk threshold.
2. According to claim 1, the method for optimizing the personalized maintenance strategy for shared rental equipment, It is characterized in that Step S1 includes the following steps: S11, collecting historical sensor data of shared rental equipment, including usage rate and usage pattern; S12, select different usage rate distribution functions for fitting and perform parameter estimation; S13, determining the optimal usage probability density function.
3. A personalized situation-based maintenance strategy optimization system for shared rental equipment, based on the method described in any one of claims 1-2, It is characterized in that Includes the following modules: Usage distribution function building module: builds the usage distribution function of the user group based on the historical sensor data of the shared rental equipment; Poisson process description module: Use Poisson process to describe the behavior of customers actively changing equipment in shared rental scenarios; Equipment degradation feature description module: Use Gamma random process and AFT model to describe the characteristics of equipment degradation and determine the mechanism of the effect of maintenance strategy on the degradation process of shared rental equipment; Detection plan establishment module: establish personalized status detection plan; Maintenance strategy establishment module: Establish the implementation methods of preventive maintenance and corrective maintenance strategies; Condition-based maintenance strategy optimization solution acquisition module: Based on the renewal theory, the average cost rate model of shared rental equipment under long-term operating conditions is obtained, and the condition-based maintenance strategy optimization solution is obtained through Monte Carlo simulation and grid search to solve the model.
