A propeller dehydrogenation process optimization method based on reliability model

Abstract

To solve the problems of the prior art, the application discloses a kind of based on reliability model's propeller hydrogen removal process optimization method, comprising: constructing the reliability model of propeller based on stress-strength interference model, strength limit and environmental stress are regarded as random variables, and the reliability is quantified by joint integral of probability density function;With the goal of maximizing reliability, an optimization model containing plating interval time, hydrogen removal heating time and hydrogen removal temperature is established, the random error influence of man-machine material ring elements is included and the constraint condition is set;GA- HC algorithm is used to solve the model, global search is realized by genetic algorithm, and local fine search is realized by climbing algorithm, and the crossover rate and mutation rate are dynamically adjusted to improve the solving efficiency. The present application aims to solve the problem that the traditional hydrogen removal process relies on experience setting and cannot quantify uncertainty, and provides a data-driven optimization scheme for high-quality manufacturing of propeller.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of reliability engineering, more particularly to a propeller dehydrogenation process optimization method based on reliability model. BACKGROUND

[0002] The propeller is the core component of the ship propulsion system, and its performance is directly related to the maneuverability, reliability and safety of the ship. In the quality control system of the propeller throughout its life cycle, the improvement of product reliability involves design optimization, precision manufacturing, quality testing and other key links. Based on the completion of structure and fluid dynamics design, the process level of the production and manufacturing link plays a decisive role in the final performance indicators of the propeller in its manufacturing process. In order to enhance the surface properties such as corrosion resistance and wear resistance, surface treatment such as electroplating is often required for key components such as blades and hubs. However, the electroplating process can cause a large amount of hydrogen to penetrate into the metal matrix, leading to hydrogen embrittlement of the material, i.e. a significant decrease in the plasticity and toughness of the material, which is prone to brittle fracture under stress, posing a serious threat to the service safety of the propeller. Therefore, after electroplating, sufficient dehydrogenation treatment must be carried out to eliminate the hydrogen embrittlement hazard. Currently, the industry generally uses heating and baking to dehydrogenate. The core parameters of the dehydrogenation process mainly include dehydrogenation temperature, holding time and time interval from electroplating completion to dehydrogenation start.

[0003] However, the existing dehydrogenation process parameter setting method mainly relies on process manuals, empirical formulas or practical experience of operators, which makes it difficult to accurately match specific product structures, material batches and equipment states, resulting in a blurred process window and low optimization degree. In addition, traditional process design regards material strength and process stress as deterministic values, without fully considering the randomness and uncertainty brought by "man-machine-material-method-environment" and other factors in actual production.

[0004] In summary, how to establish a scientific and quantitative method to comprehensively consider various uncertainty factors, optimize the dehydrogenation process in advance and systematically, and realize and ensure the target reliability required by the product is a technical problem to be solved in the field. SUMMARY

[0005] The present application provides a propeller dehydrogenation process optimization method based on reliability model to solve the problems in the prior art, which comprises the following steps:

[0006] (1) Construct a propeller reliability model R(S, E) based on the stress-strength interference model, wherein R(S, E) represents the probability that the strength limit S of the propeller is greater than the environmental stress E, and the specific process comprises,

[0007] (101) define the reliability R of the propeller as the probability P that its strength limit S is greater than the environmental stress E, i.e.

[0008] R(S,E) = P(S>E) = P(S-E>0); Equation 1

[0009] (102) Set the environmental stress and workpiece strength limit in the actual working condition of the adjustable propeller using environment as random variables, and their probability density functions are f(S) and h(E) respectively. The interference area of the two may appear S

[0010] Set the probability that the strength limit is greater than the environmental stress when the environmental stress is E0

[0011]

[0012] Where dS is the differential form of the strength limit, and the probability that the environmental stress is in the dE interval is

[0013]

[0014] Where dE is the differential form of the environmental stress;

[0015] Set S> E0 and E0-dE / 2≤E≤E0+dE / 2 as two independent events, then the probability of the two independent events occurring simultaneously is

[0016]

[0017] Where dR is the differential form of the reliability of the adjustable propeller;

[0018] Considering the entire stress interval, the reliability of the adjustable propeller can be expressed as

[0019]

[0020] Assume that the strength limit S follows a normal distribution with parameters , denoted as The environmental stress E follows a normal distribution with parameters , denoted as Set the safety margin Z = S-E, then Z also follows a normal distribution with parameters , that is

[0021]

[0022] Where Z = μ S - μ E ,

[0023] The reliability R can be calculated by the normal distribution function as

[0024]

[0025] wherein dZ is the differential form of the safety margin;

[0026] (2) Construct a propeller dehydrogenation process optimization model with the goal of maximizing reliability;

[0027] (3) Use the genetic algorithm optimized by the hill climbing algorithm to solve the propeller dehydrogenation process optimization model, and output the optimal dehydrogenation process parameters.

[0028] Further, the construction process of step (2) includes:

[0029] (201) According to formula 7, the objective function of the dehydrogenation process optimization model is:

[0030]

[0031] (202) Set the constraint condition, let be the approximate relationship model between the strength limit and the process parameters, and ε be a random error subject to normal distribution, whose mean μ ε and variance are related to the man-machine-material-loop elements, then the strength limit S is expressed as

[0032]

[0033] The strength limit range is:

[0034]

[0035] wherein S min ,S max represent the upper and lower limits of the strength limit, respectively;

[0036] Combine the process parameters and the strength limit range to construct the process optimization model.

[0037] Further, let be the approximate relationship model between the strength limit and the dehydrogenation process parameters, including the plating interval time d(h), the dehydrogenation heating time t(h), and the dehydrogenation temperature T(℃); and ε be a random error subject to normal distribution, then the range of the strength limit S is expressed as,

[0038]

[0039] The strength limit range is:

[0040]

[0041] wherein S min ,S max represent the upper and lower limits of the strength limit, respectively;

[0042] The process parameter ranges are:

[0043]

[0044] wherein d min , d max represent the upper and lower limits of the plating interval time, t min , t max represent the upper and lower limits of the hydrogen removal heating time, T min , T max represent the upper and lower limits of the hydrogen removal temperature, respectively;

[0045] The process optimization model constructed is represented by,

[0046]

[0047] d min ≤ d ≤ d max

[0048] t min ≤ t ≤ t max

[0049] T min ≤ T ≤ T max ; formula 14.

[0050] Further, the solving process of the genetic-hill climbing algorithm in step (3) includes

[0051] (301) According to the genetic algorithm, the objective function is converted into

[0052] min X F(X) = min X {1-R(X)} ; formula 15

[0053] wherein X = [d, t, T] is the process parameter, R(X) is the reliability function of the adjustable pitch propeller, and F(X) is the converted objective function;

[0054] (302) Set the dynamic adjustment of the crossover rate p c :

[0055]

[0056] wherein p c1 is the maximum crossover rate, p c2 is the minimum crossover rate, f' is the parent individual fitness, f max and f avg are the maximum fitness and average fitness of the current population, respectively;

[0057] (303) Set the dynamic adjustment of the mutation rate pm :

[0058]

[0059] Where p m1 For the maximum crossover rate, p m2 Minimum crossover rate;

[0060] (304) The genetic algorithm is optimized using the hill-climbing algorithm. The optimization process is as follows:

[0061] (3041) Define a local search neighborhood for the elite individual X0 generated by the genetic algorithm.

[0062] N(X0)={X0±Δd,X0±Δt,X0±ΔT}; Equation 18

[0063] (3042) Calculate the initial fitness F(X0). For each candidate solution X′∈N(X0) in the neighborhood, calculate F(X′). If there exists F(X′)<F(X0), then update X0=X′; otherwise, terminate the search.

[0064] This invention has at least one of the following advantages:

[0065] 1. This invention introduces a stress-strength mathematical reliability model, which models the strength limit and environmental stress in the hydrogen removal process as random variables. The reliability is quantified by combining probability density functions and integrals, which transforms the process design from the traditional "engineering experience-driven" to "data and model calculation-driven", avoiding the subjectivity of the empirical method and quantitatively evaluating the reliability of the process.

[0066] 2. This invention incorporates factors affecting quality, such as human, machine, material, and environment, into the model as random variables, quantifies the impact of random errors, and comprehensively considers the fluctuation characteristics of each factor during the optimization process, so that the obtained process parameters have stronger anti-interference ability, can effectively accommodate the uncertainties such as parameter fluctuations and environmental changes in actual production, ensure that the hydrogen removal process remains stable and reliable under complex working conditions, and improve the service robustness of controllable pitch propeller parts.

[0067] 3. This invention utilizes the global exploration capability of a genetic algorithm to cover a wide solution space, combined with the local fine-search capability of a hill-climbing algorithm to improve solution accuracy. It achieves an effective balance between global exploration and local optimization, avoiding premature convergence of the genetic algorithm or getting stuck in local optima in the hill-climbing algorithm. This significantly improves the success rate of global search for reliability optimization. At the same time, it dynamically adjusts the crossover rate and mutation rate based on the population evolution state. In the early stage, it enhances population diversity through high parameters, and in the later stage, it reduces parameters to focus on fine-tuning optimization, significantly improving solution efficiency and effectively adapting to the engineering requirements of the controllable pitch propeller hydrogen removal process. Attached Figure Description

[0068] Figure 1A flowchart of the process optimization method provided by the present application is shown.

[0069] Figure 2 A schematic diagram of calculating reliability by joint integration of the probability density function of the strength limit and the environmental stress is shown.

[0070] Figure 3 A flowchart of the GA-HC algorithm solving process provided by the present application is shown.

[0071] Figure 4 A schematic diagram of the mean square error change process in the approximate relationship model training process between the strength limit and the process parameters is shown.

[0072] Figure 5 A schematic diagram of the fitness change in the GA-HC algorithm iteration operation process provided by the present application is shown. DETAILED DESCRIPTION

[0073] The present application will now be further described in detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams and only show the basic structure of the present application in a schematic manner, and therefore only show the components related to the present application.

[0074] Please note that the "upper", "lower", "left", "right", "top end", "top", "bottom end", "bottom" and the like used by the present application to describe the positional relationship do not represent the absolute positional relationship between the modules / components / elements / units / parts, but the relative positional relationship between the modules / components / elements / units / parts.

[0075] As Figure 1 shown, the present application exemplarily provides a propeller hydrogen removal process optimization method based on a reliability model, including the following steps:

[0076] Step one, build a propeller reliability model R(S, E) based on a stress-strength interference model; the reliability model based on the stress-strength interference model is the probability that the strength limit S of the propeller is greater than the environmental stress E, as Figure 2 shown.

[0077] Preferably, the strength limit of the product is S, the use environmental stress is E, and the propeller reliability R under normal working conditions of the propeller is the probability P that the strength limit S is greater than the environmental stress E, which is expressed as

[0078] R(S, E) = P(S > E) = P(S - E > 0); Equation 1

[0079] The environment stress and the workpiece strength limit in the actual working condition of the pitch control propeller are both random variables, and their probability density functions are f(S) and h(E) respectively. The interference area of the two may appear S < E, which leads to the failure of the pitch control propeller;

[0080] When the environment stress is E0, the probability that the strength limit is greater than the environment stress is

[0081]

[0082] The probability that the environment stress is in the dE interval is

[0083]

[0084] Where dE is the differential form of the environment stress;

[0085] When S > E0 and E0-dE / 2 ≤ E ≤ E0+dE / 2 are two independent events, the probability that the two independent events occur simultaneously is

[0086]

[0087] Where dR is the differential form of the reliability of the pitch control propeller;

[0088] Considering the entire stress interval, the reliability of the pitch control propeller can be expressed as

[0089]

[0090] Suppose that the strength limit S obeys the normal distribution with parameters , and is denoted as The environment stress E obeys the normal distribution with parameters , and is denoted as Set the safety margin Z = S-E, and then Z also obeys the normal distribution with parameters , that is

[0091]

[0092] Where μ Z = μ S - μ E ,

[0093] The reliability model of the pitch control propeller represented by the normal distribution function is

[0094]

[0095] Where dZ is the differential form of the safety margin.

[0096] Step two, build the optimization model of the propeller hydrogen removal process; the optimization model of the propeller hydrogen removal process is a hydrogen removal process parameter optimization model with the maximum reliability of the propeller as the target.

[0097] Preferably, the propeller reliability model R(S, E) built in the previous step is used as the optimization target, and the maximization of R(S, E) is pursued, and the objective function is represented as

[0098]

[0099] The strength limit of the propeller after the hydrogen removal process introduces the random error influence of the man-machine-material-loop elements, wherein the process parameters selected are the plating interval time d(h), the hydrogen removal heating time t(h) and the hydrogen removal temperature T(℃), and are represented as

[0100]

[0101] is an approximate relationship model between the strength limit and the key parameters of the hydrogen removal process, including the plating interval time d(h), the hydrogen removal heating time t(h) and the hydrogen removal temperature T(℃), which is calculated from historical test data, and ε is a random error, which obeys a normal distribution with a mean of 0 and a variance of ; and the mean μ ε , the variance and the influence relationship of the man-machine-material-loop elements are represented as

[0102]

[0103] wherein x i represents the i-th man-machine-material-loop element, is the mean of x i ;

[0104] Considering the process requirements, the process parameters and the strength limit need to be within the process specification range during the optimization process, that is, the constraint condition is set as

[0105]

[0106] wherein S min , S max respectively represent the upper and lower limits of the strength limit, d min , d max respectively represent the upper and lower limits of the plating interval time, t min , t max respectively represent the upper and lower limits of the hydrogen removal heating time, and T min , T max ​respectively represent the upper and lower limits of the temperature of hydrogen removal. The approximate relationship model between the strength limit S and the plating interval time d(h), the hydrogen removal heating time t(h) and the hydrogen removal temperature T(℃) is obtained by BP neural network nonlinear fitting according to the historical data of a certain controllable pitch propeller The mean square error change process in the training process is shown in Figure 4 .

[0107] The process optimization model constructed is represented as

[0108]

[0109] d min ≤d≤d max

[0110] t min ≤t≤t max

[0111] T min ≤T≤T max ; equation 14

[0112] Step three, the GA-HC algorithm is used to solve the hydrogen removal process parameter optimization model in the last step; the GA-HC algorithm is a genetic algorithm (Genetic Algorithm, GA) optimized by using the hill-climbing algorithm (Hill-Climbing, HC).

[0113] Preferably, the GA-HC algorithm performs global search through GA to explore the solution space, and combines the local fine search capability of HC to improve the accuracy of the solution, the algorithm process is shown in Figure 3 , and the core solving idea can be represented as

[0114] min X F(X)=min X {1-R(X)}; equation 15

[0115] Where X=[d,t,T] is the process parameter, R(X) is the reliability function of the controllable pitch propeller, and F(X) is the transformed objective function;

[0116] The crossover rate p c is set in the solving process

[0117]

[0118] Where p c1 is the maximum crossover rate, p c2 is the minimum crossover rate, f′ is the fitness of the parent individual, f max and f avg are the maximum fitness and average fitness of the current population, respectively.

[0119] Setting mutation rate p in solving process c Adjustment process is expressed as dynamic adjustment of population individual fitness

[0120]

[0121] Where p m1 is the maximum crossover rate, p m2 is the minimum crossover rate, f' is the parent individual fitness, f max and f avg are the maximum fitness and average fitness of the current population respectively.

[0122] The specific content of optimizing GA by HC preferably includes:

[0123] Setting local search field for elite individual X0 generated by GA

[0124] N(X0) = {X0 ± Δd, X0 ± Δt, X0 ± ΔT}; Equation 18

[0125] Calculating initial fitness F(X0); calculating F(X') for each candidate solution X' in the field N(X0); if F(X') < F(X0) exists, updating X0 = X', otherwise terminating local search.

[0126] Embodiment 2

[0127] In this embodiment, the constraint conditions are respectively 0 ≤ d ≤ 4h for plating interval time, 12h ≤ t ≤ 21h for hydrogen removal time, and 180℃ ≤ T ≤ 200℃ for temperature; considering the influence of man-machine-material-environment elements, the influence of man-machine-material-environment is added to the strength limit S in the form of random error ε, and the random error ε obeys normal distribution with mean value of 0 and variance of 5, recorded as ε ~ N(0, 5), and the strength limit of the pitch prop after considering the random error cannot exceed its process requirement S ≤ 1050MPa, so the hydrogen removal process parameter optimization process is:

[0128] 1. Constructing a reliability model R(S, E) of the pitch prop based on the stress-strength interference model; the reliability model of the hydrogen removal process based on the stress-strength interference model is the probability that the strength limit S of the pitch prop is greater than the environmental stress E, S, as shown in the following formula: Figure 2

[0129] 2. Constructing a pitch prop hydrogen removal process optimization model; the pitch prop hydrogen removal process optimization model is a hydrogen removal process parameter optimization model with the maximum reliability of the pitch prop as the target; the optimization variables are plating interval time d (h), hydrogen removal heating time t (h), and hydrogen removal temperature T (℃), and according to the above constraint conditions, the process optimization model can be expressed as:

[0130] ​

[0131] 3. Using GA-HC algorithm to solve the hydrogen removal process parameter optimization model in the last step; further local search on the elite individuals obtained by GA algorithm through HC algorithm; setting the maximum iteration number in the algorithm as 100, the population size as 100, the initial crossover rate p c =0.9 and the initial mutation rate p m =0.1, and the chromosome individual adopts decimal encoding mode.

[0132] The fitness change of the algorithm iteration operation is shown in Figure 5 The fitness value basically remains unchanged after the algorithm iteration to the 23rd time, at this time the maximum fitness is 99.927, at this time the corresponding hydrogen removal temperature is 193℃, the hydrogen removal time is 15h, and the plating interval time is 2.5h.

[0133] Based on the above ideal embodiments according to the present application, through the above description, relevant staff can make various changes and modifications without deviating from the technical idea of the present application. The technical scope of the present application is not limited to the content in the specification, and the technical scope must be determined according to the scope of claims.

Claims

1. A method for propeller dehydrogenation process optimization based on reliability model, characterized in that, Comprise the following steps: (1) Construct the reliability model R(S, E) of the controllable pitch propeller based on the stress-strength interference model, wherein the R(S, E) represents the probability that the strength limit S of the controllable pitch propeller is greater than the environmental stress E, and the specific process comprises: (101) Defining the reliability R of the controllable pitch propeller as the probability P that the strength limit S thereof is greater than the environmental stress E, that is, R(S, E) = P(S > E) = P(S - E > 0); Equation 1 (102) Assuming that the environmental stress and the strength limit of the workpiece of the controllable pitch propeller in the actual working condition are both random variables, the probability density functions thereof are f(S) and h(E) respectively, and the interference region of the two may have the case that S < E, which leads to the failure of the controllable pitch propeller; Assuming that when the environmental stress is E0, the probability that the strength limit is greater than the environmental stress is wherein dS is the differential form of the strength limit, and the probability that the environmental stress is in the interval dE is wherein dE is the differential form of the environmental stress; Assuming that S > E0 and E0 - dE / 2 ≤ E ≤ E0 + dE / 2 are two independent events, the probability that the two independent events occur simultaneously is wherein dR is the differential form of the reliability of the controllable pitch propeller; Considering the case of the entire stress interval, combining the joint integral of the probability density functions, the reliability of the controllable pitch propeller can be represented as Suppose that the strength limit S follows a normal distribution with parameters , denoted as Suppose that the environmental stress E follows a normal distribution with parameters , denoted as Set the safety margin Z = S - E, then Z also follows a normal distribution with parameters , i.e. where μ Z = μ S - μ E , Through the calculation of the normal distribution function, the reliability R can be obtained as wherein dZ is the differential form of the safety margin; (2) Constructing a controllable pitch propeller hydrogen removal process optimization model with the maximum reliability as the target; (3) Solving the controllable pitch propeller hydrogen removal process optimization model by using the genetic algorithm optimized by the hill climbing algorithm, and outputting the optimal hydrogen removal process parameters.

2. The method of claim 1, wherein, The construction process of step (2) comprises: (201) According to Equation 7, the objective function of the hydrogen removal process optimization model is obtained as: (202) Set constraints, set is an approximate relationship model between the ultimate strength and the process parameters, ε is a random error obeying normal distribution, whose mean μ ε and variance is related to the man-machine-material elements, and the strength limit S is expressed as Then the strength limit range can be obtained as: where S min , S max denote the upper and lower limits of the strength limit, respectively; Combining the process parameters and the strength limit range, the process optimization model is constructed.

3. The controllable pitch propeller hydrogen removal process optimization method based on the reliability model according to claim 2, wherein Let is an approximate relationship model between the strength limit and the plating interval time d (h), the hydrogen removal heating time t (h), and the hydrogen removal temperature T (°C); ε is a random error subject to a normal distribution, and the range of the strength limit S is represented as, Then the strength limit range can be obtained as: where S min , S max represent the upper and lower limits of the strength limit, respectively; The process parameter range is: wherein d min , d max represent the upper and lower limits of the plating interval time, respectively, t min , t max represent the upper and lower limits of the hydrogen removal heating time, respectively, T min , T max represent the upper and lower limits of the hydrogen removal temperature, respectively; Then the constructed process optimization model is represented as, 4. The method of claim 3, wherein, The solving process of the genetic-hill climbing algorithm in step (3) comprises (301) According to the genetic algorithm, the objective function is converted into min X F(X) = min X {1 - R(X)} ; Equation 15 wherein X = [d, t, T] is the process parameter, R(X) is the reliability function of the controllable pitch propeller, and F(X) is the converted objective function; (302) Set dynamic adjustment of cross rate p c : where p c1 is the maximum crossover rate, p c2 is the minimum crossover rate, f′ is the fitness of the parent individual, f max and f avg are the maximum and average fitness of the current population, respectively. (303)Setting a dynamic adjustment of the variability rate p m : where p m1 is the maximum cross rate, p m2 is the minimum cross rate; (304) Using the hill climbing algorithm to optimize the genetic algorithm, and the optimization process is: (3041) Defining the local search field of the elite individual X0 generated by the genetic algorithm as N(X0) = {X0 ± Δd, X0 ± Δt, X0 ± ΔT}; Equation 18 (3042) Calculating the initial fitness F(X0), calculating F(X') for each candidate solution X' ∈ N(X0) in the field, if there is F(X') < F(X0), updating X0 = X', otherwise terminating the search.

Images