Method and device for determining water pipe scaling thickness based on data center machine room

Abstract

The invention discloses a water pipe scaling thickness determination method and device based on a data center machine room, and relates to the field of artificial intelligence. The method comprises the following steps: taking water hardness information, total dissolved solid information, pH value information, water flow velocity information, water temperature information and collection time information collected from a water pipe as model input variables; according to prior knowledge learned by a fuzzy neural network model in a model training stage, a water pipe scaling thickness at a target moment is determined according to a model input variable, and a training target of the fuzzy neural network model comprises: according to a training sample set and a network parameter set, and adjusting the membership degree and the weight parameter of the fuzzy neural network by using a starfish optimization algorithm until the loss function value of the fuzzy neural network reaches a preset threshold. The technical problems that in an existing pipeline scaling prediction method, parameter coupling optimization is difficult, local convergence is prone to falling into a local optimal solution, and consequently pipeline scaling thickness detection is inaccurate are solved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence, and more specifically, to a method and device for determining the thickness of scale buildup in water pipes in a data center computer room. Background Art

[0002] In a bank's data center, scaling problems caused by water quality in the cooling system pipes have become increasingly prominent. This type of deposit not only reduces heat exchange efficiency but also poses a risk of equipment failure and downtime. Against the backdrop of the rapid iteration of intelligent technologies, data-driven modeling techniques have opened up new avenues for predicting the status of industrial systems. Traditional machine learning methods are limited in their ability to extract time-varying features, and when applied to such scenarios, they still have many limitations, making it difficult to meet practical needs. In recent years, thanks to the rapid iteration of computing hardware and breakthroughs in sensing technology, neural network models combined with intelligent optimization strategies have become a research direction of widespread interest in academia and industry.

[0003] However, existing pipeline scaling prediction methods still have the problem of parameter coupling optimization and local convergence easily falling into the local optimal solution, resulting in inaccurate technical problems in the thickness detection of pipeline scaling.

[0004] To address the above-mentioned problems, no effective solutions have been proposed so far. Summary of the Invention

[0005] The embodiments of the present application provide a method and device for determining the thickness of water pipe scale in a data center computer room, so as to at least solve the technical problems in existing pipe scale prediction methods, such as the difficulty of parameter coupling optimization and the local convergence that easily falls into the local optimal solution, resulting in inaccurate detection of pipe scale thickness.

[0006] According to one aspect of an embodiment of the present application, a method for determining the thickness of scale in a water pipe is provided, comprising: using water hardness information, total dissolved solids information, pH information, water flow rate information, water temperature information, and collection time information collected from the water pipe as model input variables; determining the thickness of scale in the water pipe at a target moment based on the model input variables using the prior knowledge learned by the fuzzy neural network model during the model training phase, wherein the training objectives of the fuzzy neural network model include: adjusting the membership and weight parameters of the fuzzy neural network using the starfish optimization algorithm according to the training sample set and the network parameter set, until the loss function value of the fuzzy neural network reaches a preset threshold.

[0007] Optionally, the training process of the fuzzy neural network model includes the following steps: initializing a network parameter set of the fuzzy neural network, wherein the network parameter set includes at least: the center value of the membership function, the width of the membership function, and the weight; based on the network parameter set and the training sample set after successful initialization, using the starfish optimization algorithm to adjust the membership and weight parameters of the fuzzy neural network until the loss function value of the fuzzy neural network reaches a preset threshold, and the fuzzy neural network model is obtained by training.

[0008] Optionally, the fuzzy neural network model includes an input layer, a fuzzification layer, a rule layer, a normalization layer and an output layer, wherein: the input layer is used to input the model input variables into the fuzzy neural network model, wherein the number of nodes in the input layer is equal to the number of dimensions of the model input variables; the fuzzification layer is used to fuzzify the model input variables and map the input space to a Gaussian distribution space, wherein the number of nodes in the fuzzification layer is equal to the number of fuzzy levels of the model neural network model; each node of the fuzzification layer represents a fuzzy verbal variable value and is used to calculate the membership function of the components of the input vector to the fuzzy set of each verbal variable; the rule layer is used to calculate the total fuzzy membership of the model input variables through fuzzy rules, wherein the number of summary points in the rule layer is equal to the number of fuzzy rules, and each node in the rule layer is used to determine the applicability of a fuzzy rule; the normalization layer is used to perform normalized calculation on the applicability of each fuzzy rule, wherein the total number of nodes in the normalization layer is equal to the total number of nodes in the rule layer; the output layer is used to integrate fuzzy rules to determine the output result of the fuzzy neural network model.

[0009] Optionally, the execution of the starfish optimization algorithm goes through the following stages: an initialization stage, which is used to generate a position initialization matrix, evaluate the fitness values ​​of all individuals in the population through an objective function, and construct a matrix for storing and updating fitness values ​​based on the position initialization matrix; a selection stage, which is used to select to enter the exploration stage or the development stage based on the comparison result of the random number and the global stage selection threshold; the exploration stage, which is used to update the position of each individual in the population using a one-dimensional search mode, wherein the one-dimensional search mode is used to update the position of the current individual by simulating the exploration behavior of the starfish tentacles and using the position information of other individuals in the population except the current individual; the development stage, which is used to search for the global optimal solution by simulating the predation or regeneration behavior of the starfish; the iterative cycle stage, which is used to update the fitness values ​​of all individuals in the population after each iteration and store the current optimal solution; if the population does not meet the termination condition after any iteration, it returns to the selection stage to start a new cycle; if the population meets the termination condition after any iteration, the optimal solution obtained in the last round is retained as the final output result, wherein the final output result is used to optimize the parameters of the fuzzy neural network.

[0010] Optionally, a parallel bidirectional search strategy is used to use the distance parameter between two random individuals in the population as an update basis to control some candidate solutions of the population to move in a better direction, while other candidate solutions in the population except this part of the candidate solutions perform reverse movement within the same iteration cycle; the regeneration strategy is used to update the position of the individual with the smallest fitness value in the population by moving at a preset speed to avoid the individual being determined as the local optimal solution.

[0011] Optionally, a one-dimensional search mode is used in the exploration phase to update the position of each individual in the population, including: detecting the position information of the i-th individual in the population at the t-th iteration, where the i-th individual is any individual in the population and t is an integer greater than or equal to 1; detecting the position information of any two randomly selected individuals in the population; updating the position information of the i-th individual based on the position information of the i-th individual in the population at the t-th iteration, the position information of any two randomly selected individuals in the population, and a preset random number.

[0012] Optionally, the development phase searches for the global optimal solution by simulating the predation behavior of starfish, including: updating the position information of each individual in the population according to the position information of the i-th individual in the population at the t-th iteration, the position information of the j-th individual at the t-th iteration, and the current global optimal solution of the population at the t-th iteration, wherein the i-th individual is any individual in the population, and the j-th individual is an individual randomly selected from the population except the i-th individual; determining the global optimal solution of the population after the t-th iteration according to the updated position information of each individual in the population at the t-th iteration.

[0013] Optionally, the development phase searches for the global optimal solution by simulating the regeneration behavior of starfish, including: determining the movement speed of the i-th individual at the t+1-th iteration based on the position information of the i-th individual in the population at the t-th iteration, the position information of the individual with the smallest fitness value in the population at the t-th iteration, and a preset minimum value; and determining the global optimal solution of the population at the t+1-th iteration based on the movement speed of each individual in the population at the t+1-th iteration.

[0014] According to another aspect of the present application, a computer-readable storage medium is provided, comprising: a computer program stored in the computer-readable storage medium, wherein when the computer program is executed, the device where the computer-readable storage medium is located executes the above-mentioned method for determining the thickness of scale in water pipes.

[0015] According to another aspect of the present application, an electronic device is also provided, comprising one or more processors and a memory, wherein the memory is used to store one or more programs, wherein when the one or more programs are executed by one or more processors, the one or more processors execute the above-mentioned method for determining the thickness of scale in a water pipe.

[0016] According to another aspect of the present application, a computer program product is further provided, wherein the computer program product includes a computer program or instructions, and when the computer program or instructions are executed by a processor, the computer program or instructions implements the above-mentioned method for determining the scale thickness of water pipes based on a data center computer room.

[0017] From the above content, it can be seen that the present application uses the water hardness information, total dissolved solids information, pH information, water flow rate information, water temperature information and collection time information collected from the water pipe as model input variables; through the prior knowledge learned by the fuzzy neural network model in the model training stage, the scale thickness of the water pipe at the target time is determined based on the model input variables, wherein the training objectives of the fuzzy neural network model include: according to the training sample set and the network parameter set, using the starfish optimization algorithm to adjust the membership and weight parameters of the fuzzy neural network until the loss function value of the fuzzy neural network reaches the preset threshold.

[0018] In the embodiments of the present application, the present application introduces the starfish optimization algorithm to globally optimize the parameters of the fuzzy neural network, thereby overcoming the defect that the traditional method is prone to falling into local optimality, thereby improving the model's prediction accuracy for the thickness of scale buildup in the water pipes in the computer room, and utilizing the search strategy of the starfish optimization algorithm, combined with the fuzzy rule adaptive adjustment mechanism, to achieve the global optimal configuration of the fuzzy neural network parameters, solving the parameter coupling optimization problem in the existing methods. Secondly, the introduction of the fuzzy neural network can effectively handle the fuzzy relationship between water quality parameters, and combined with the global optimization capability of the starfish optimization algorithm, it can better model the complex nonlinear relationship between scale thickness and water quality parameters. Therefore, according to the technical solution of the present application, the optimized fuzzy neural network model can provide an accurate prediction basis for the water quality management of the computer room, thereby optimizing the operating efficiency of the cooling system, extending the operating stability of the computer room, and reducing the risk of equipment failure and downtime due to scaling of the water pipes. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0020] Figure 1 is a flow chart of an optional method for determining the thickness of scale in a water pipe according to an embodiment of the present application;

[0021] Figure 2 is a flow chart of an optional fuzzy neural network according to an embodiment of the present application;

[0022] Figure 3 is a flowchart of an optional starfish optimization algorithm according to an embodiment of the present application;

[0023] Figure 4 is an overall flow chart of an optional method for determining the thickness of water pipe scale according to an embodiment of the present application;

[0024] Figure 5 This is a schematic diagram of an optional device for determining the thickness of scale in a water pipe according to an embodiment of the present application. DETAILED DESCRIPTION

[0025] In order to enable those skilled in the art to better understand the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of this application.

[0026] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0027] It should also be noted that the information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) collected by this application are information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of relevant data comply with the relevant laws, regulations and standards of the relevant regions, take necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation entrances for users to choose to authorize or refuse. For example, an interface is set up between this system and relevant users or institutions. Before obtaining relevant information, it is necessary to send an acquisition request to the aforementioned user or institution through the interface, and obtain relevant information after receiving the consent information fed back by the aforementioned user or institution.

[0028] According to an embodiment of the present application, an embodiment of a method for determining the thickness of scale in a water pipe is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0029] Optionally, according to an embodiment of the present application, a water pipe scale thickness detection system (hereinafter referred to as the system) is provided as the execution subject of the method for determining the water pipe scale thickness in the embodiment of the present application, wherein the system can be a software system or an embedded system combining software and hardware. Of course, the method execution subject in the embodiment of the present application can also be other forms of execution subjects, such as devices, equipment, etc. Those skilled in the art should know that this application does not specifically limit the specific form of expression of the method execution subject.

[0030] Figure 1 is a flow chart of an optional method for determining the thickness of water pipe scale according to an embodiment of the present application, such as Figure 1 As shown, the method includes the following steps:

[0031] In step S101, the water hardness information, total dissolved solids information, pH value information, water flow rate information, water temperature information and collection time information collected from the water pipe are used as model input variables.

[0032] Alternatively, water hardness refers to the content of divalent metal ions, such as calcium and magnesium, in water and is a key factor influencing scaling in water pipes. Water with high hardness is more likely to form mineral deposits within pipes, leading to scaling. In this application, water hardness is used as one of the input variables to train a fuzzy neural network to identify its impact on scale thickness.

[0033] Optionally, total dissolved solids (TDS) refers to the total content of various organic and inorganic salts dissolved in water, including but not limited to calcium, magnesium, sodium, potassium, sulfate, chloride, etc., which have a direct impact on scale formation. TDS measurements are also included as input variables in this application to help the model more comprehensively assess scaling risk.

[0034] Alternatively, pH is an indicator of the acidity and alkalinity of water. Water with different pH values ​​has a significant impact on the rate and type of scale formation. In this application, pH is used as an input variable to train the model to identify its impact on scaling.

[0035] Alternatively, water flow velocity has a significant impact on scale formation: higher flow velocities can reduce scale formation, while lower flow velocities promote sediment accumulation. This application uses water flow velocity information as one of the input variables, aiming to predict scaling trends at different flow rates using a fuzzy neural network model.

[0036] Alternatively, water temperature is another key factor influencing scale formation. Water at different temperatures has different chemical balances, which affects the rate and type of scaling. Collecting water temperature information and incorporating it into input variables can help the model more accurately predict scaling.

[0037] Alternatively, time is an indispensable variable in scaling prediction, as scaling is a process that accumulates over time. By recording water quality parameters at different time points, this application aims to train a model to predict scaling at a certain point in the future.

[0038] Optionally, the system feeds data into a fuzzy neural network, passing the input values ​​to the next layer. The input layer ensures that the fuzzy neural network model accurately receives and processes data that matches the dimensions of the input variables. This ensures that water quality parameters such as hardness, TDS, and pH are directly and fully utilized by the model, providing a solid foundation for subsequent fuzzification, rule calculation, and prediction output, thereby improving the accuracy and reliability of the model's predictions of water pipe scale thickness.

[0039] Step S102 , using the prior knowledge learned by the fuzzy neural network model during the model training phase, the water pipe scale thickness at the target time is determined based on the model input variables.

[0040] In step S102, the training objectives of the fuzzy neural network model include: adjusting the membership and weight parameters of the fuzzy neural network using the starfish optimization algorithm according to the training sample set and the network parameter set, until the loss function value of the fuzzy neural network reaches a preset threshold.

[0041] Optionally, the target moment may refer to a scenario where the thickness of scale in a water pipe needs to be predicted at a certain point in the future. This is an application scenario where, after the model training is completed, the water quality parameters at the current moment and the time point to be predicted are input to predict the thickness of scale at a specific moment in the future.

[0042] Optionally, the Starfish Optimization Algorithm (STARFI) is a heuristic search algorithm that simulates the feeding and regeneration behavior of the marine organism starfish, searching for an optimal solution in parameter space. The system uses the Starfish Optimization Algorithm to optimize the membership and weight parameters of the fuzzy neural network, enabling the model to more accurately reflect the complex relationship between water quality parameters and scale thickness.

[0043] Optionally, the fuzzy neural network is trained to minimize a loss function by iteratively optimizing parameters until the loss function reaches a preset threshold. This threshold is chosen based on the required prediction accuracy to ensure that the model achieves the expected prediction performance after training.

[0044] From the above content, it can be seen that the present application introduces the starfish optimization algorithm to globally optimize the parameters of the fuzzy neural network, thereby overcoming the defect that the traditional method is prone to falling into local optimality, thereby improving the model's prediction accuracy for the thickness of scale buildup in the water pipes in the computer room, and utilizing the search strategy of the starfish optimization algorithm, combined with the fuzzy rule adaptive adjustment mechanism, to achieve the global optimal configuration of the fuzzy neural network parameters, solving the parameter coupling optimization problem in the existing methods. Secondly, the introduction of the fuzzy neural network can effectively handle the fuzzy relationship between water quality parameters, and combined with the global optimization capability of the starfish optimization algorithm, it can better model the complex nonlinear relationship between scale thickness and water quality parameters. Therefore, according to the technical solution of the present application, the optimized fuzzy neural network model provides an accurate prediction basis for the water quality management of the computer room, thereby optimizing the operating efficiency of the cooling system, extending the operating stability of the computer room, and reducing the risk of equipment failure and downtime due to water pipe scaling.

[0045] In an optional embodiment, the training process of the fuzzy neural network model includes the following steps: initializing a network parameter set of the fuzzy neural network, wherein the network parameter set includes at least: the center value of the membership function, the width of the membership function, and the weight; based on the network parameter set and the training sample set after successful initialization, using the starfish optimization algorithm to adjust the membership and weight parameters of the fuzzy neural network until the loss function value of the fuzzy neural network reaches a preset threshold, and the fuzzy neural network model is trained.

[0046] Optionally, the network parameter set can be expressed as c=(c 11 ,c 21 ,…,c i1 ,c 12 ,c 22 ,…,c ij )σ=(σ 11 ,σ 21 ,…,σ i1 ,σ 12 ,σ 22 ,…,σ ij )w=(w 1 ,w 2 ,…,w k ), the input sample set can be expressed as X=[X1,X2,…,X n ].

[0047] Optionally, during the initialization phase, the system can set multiple sets of fuzzy neural network parameters—namely, the membership function center value c, the membership function width σ, and the weight w—to X1, X2, and X3, respectively, and establish a matrix X. Each parameter combination is considered a starfish individual, located at a specific position in the multidimensional parameter space. After generating the position initialization matrix, the fitness values ​​of all individuals in the population can be obtained through objective function evaluation.

[0048] Calculating the fitness value (typically accomplished through an objective function such as the root mean square error (RMSE)) is key to evaluating the performance of each individual. It not only measures the model's prediction accuracy under the current parameter configuration but also indirectly indicates which areas in the parameter space warrant further exploration. The Starfish Optimization algorithm leverages this information to iteratively update the positions of individuals within the population, gradually approaching the optimal solution. Specifically, individuals with lower fitness values ​​(i.e., smaller RMSE values) indicate better model performance.

[0049] In addition, the system can set up a matrix F (dimension N×1) to store and update fitness values. This is used to organize the storage and management of fitness values, facilitating real-time comparison and tracking of changes in the best individuals during the iteration process. As the algorithm iterates, individuals with lower fitness values ​​in the F matrix are retained, while those with poor performance are eliminated or updated. This process ensures that the population as a whole evolves towards higher predictive performance. After the system is initialized, the Starfish Optimization Algorithm can enter the main loop and begin iterative calculations in the exploration and development phases.

[0050] In summary, through random initialization of parameters and dynamic adjustment based on fitness values, the Starfish optimization algorithm effectively guides the training process of the fuzzy neural network model and promotes the optimization of model parameters, thereby significantly improving the accuracy of the prediction of scale thickness of the water pipes in the computer room and ensuring the high performance and stability of the model.

[0051] In an optional embodiment, the fuzzy neural network model includes an input layer, a fuzzification layer, a rule layer, a normalization layer and an output layer, wherein:

[0052] The input layer is used to input the model input variables into the fuzzy neural network model, where the number of nodes in the input layer is equal to the number of dimensions of the model input variables;

[0053] The fuzzification layer is used to fuzzify the model input variables and map the input space to the Gaussian distribution space. The number of nodes in the fuzzification layer is equal to the number of fuzzy levels of the model neural network model. Each node in the fuzzification layer represents a fuzzy verbal variable value and is used to calculate the membership function of the components of the input vector to the fuzzy set of each verbal variable.

[0054] The rule layer is used to calculate the total fuzzy membership of the model input variables through fuzzy rules. The number of summary points in the rule layer is equal to the number of fuzzy rules, and each node in the rule layer is used to determine the applicability of a fuzzy rule.

[0055] The normalization layer is used to normalize the applicability of each fuzzy rule, where the total number of nodes in the normalization layer is equal to the total number of nodes in the rule layer;

[0056] The output layer is used to integrate fuzzy rules to determine the output results of the fuzzy neural network model.

[0057] Optionally, Figure 2 is a flow chart of an optional fuzzy neural network according to an embodiment of the present application, such as Figure 2 As shown in Figure 1, the first layer of the fuzzy neural network is the input layer: the input layer receives the vector (X) of six independent parameters, inputs the data into the fuzzy neural network, and transmits the input value to the next layer, which can be expressed as formula (1), as shown below:

[0058] X=(x1,x2,…,x6) T (1)

[0059] Among them, X is the input vector, x1 is the water hardness, x2 is the total dissolved solids (TDS), x3 is the pH value, x4 is the water flow rate, x5 is the water temperature, and x6 is the time. The number of nodes in this layer is 6.

[0060] Optionally, the input layer ensures that the fuzzy neural network model can accurately receive and process data that matches the dimensions of the input variables. This ensures that water quality parameters such as hardness, TDS, and pH can be directly and completely utilized by the model, providing a solid foundation for subsequent fuzzification, rule calculation, and prediction output, thereby improving the accuracy and reliability of the model's prediction of water pipe scale thickness.

[0061] The second layer is the fuzzification layer: fuzzy processing is performed on the data, and the input space is mapped to the Gaussian distribution space. The number of nodes is K i is relative to x i Each node represents a fuzzy language variable value, and its function is to calculate the membership function of the input vector components to the fuzzy set of each language variable. Where K is the number of levels of fuzzy rules. There are K groups of membership functions, and each group has P membership functions. The membership function can be expressed as formula (2):

[0062]

[0063] Among them, c ij is the central value of the membership function, σ ij is the width of the membership function, and t is the number of iterations of the starfish optimization algorithm.

[0064] The third layer is the rule layer: the data is represented by fuzzy rules to calculate the total fuzzy membership of the data. The total number of points in this layer is k, which is used to judge the applicability of each rule. The output is shown in formula (3):

[0065]

[0066] where u k (t) is the output value of the kth neuron.

[0067] The fourth layer is the normalization layer: the applicability of each rule is normalized. The number of nodes in the normalization layer is consistent with that in the rule layer. The output is shown in formula (4):

[0068]

[0069] Among them, v k (t) is the output value of the k-th neuron (expressed in normalized form).

[0070] The fifth layer is the output layer: fuzzy rules are integrated to achieve clear calculation output as shown in formula (5):

[0071]

[0072] Among them, y(t) is the output of the fuzzy neural network, w k (t) is the weight of the kth normalized neuron in the fuzzy neural network.

[0073] In an optional embodiment, the Starfish optimization algorithm is executed through the following stages:

[0074] The initialization phase is used to generate a position initialization matrix, evaluate the fitness values ​​of all individuals in the population through the objective function, and construct a matrix for storing and updating fitness values ​​based on the position initialization matrix;

[0075] The selection phase is used to select whether to enter the exploration phase or the development phase based on the comparison result of the random number and the global phase selection threshold;

[0076] The exploration phase is used to update the position of each individual in the population using a one-dimensional search mode. The one-dimensional search mode is used to update the position of the current individual by simulating the exploration behavior of starfish tentacles and using the position information of other individuals in the population except the current individual.

[0077] The development phase is used to search for the global optimal solution by simulating the predation or regeneration behavior of starfish;

[0078] The iterative cycle stage is used to update the fitness values ​​of all individuals in the population and store the current optimal solution after each iteration. If the population does not meet the termination condition after any iteration, it returns to the selection stage and starts a new cycle. If the population meets the termination condition after any iteration, the optimal solution obtained in the last round is retained as the final output result, where the final output result is used to optimize the parameters of the fuzzy neural network.

[0079] Optionally, Figure 3 is a flow chart of an optional starfish optimization algorithm according to an embodiment of the present application, such as Figure 3 As shown in Figure 1, first, the algorithm starts and prepares for the optimization process. Then, the system initializes the population and creates a set of initial positions of multiple starfish individuals. Each individual represents a set of fuzzy neural network parameter configurations. The system can set the parameters of multiple sets of fuzzy neural networks, namely the center value c of the membership function, the width σ of the membership function, and the weight w, to X 1 、X 2 and X 3 , and establish the matrix X. After generating the position initialization matrix, the fitness values ​​of all individuals in the population can be obtained through objective function evaluation. The fitness value is evaluated by the objective function and reflects the quality of the individual (parameter configuration). In the starfish optimization algorithm, it is used to identify which starfish individuals (fuzzy neural network parameter combinations) are closer to the optimal solution to the problem. The fitness vector is shown in Equation (6):

[0080]

[0081] in, represents the spatial position of the jth starfish individual in the i-th dimension (i=1,2,3), rand is a random number in the interval (0,1), u j and l j represent the upper and lower bounds of the i-th dimension design variable respectively.

[0082] In addition, the system can set up a matrix F (dimension N×1) to store and update fitness values. After the system is initialized, the starfish optimization algorithm can enter the main loop process and start executing iterative calculations in the exploration phase and the development phase. As shown in formula (7):

[0083]

[0084] Among them, the fitness function F(X i ) is the value of X i The root mean square error (RMSE) in the training set directly reflects the average deviation between the model's predicted values ​​and the actual values. A lower RMSE value indicates a more accurate model prediction.

[0085] Next, the system can use a stage selection mechanism to decide whether the current iteration enters the exploration stage or the development stage. This process depends on the random number rand generated in the (0,1) interval and the algorithm control parameter G p (Global stage selection threshold, default value G p =0.5). <G p Enter the exploration phase when rand>G p Entering the development phase.

[0086] During the exploration phase, the system determines the direction and magnitude of an individual's movement by calculating the direction angle θ and the energy variable Et. The calculation of θ and Et is based on the current number of iterations T, the maximum number of iterations Tmax, and the algorithm parameters, which together determine the range and depth of the individual's exploration.

[0087] In addition, during the exploration phase, each starfish individual (fuzzy neural network parameter configuration) can update its position in the parameter space based on the calculated direction angle θ and energy variable Et through Equation (8), that is, adjust the membership function and weight parameters. This update strategy imitates the expansion and contraction of the starfish tentacles in a single-dimensional space, which is equivalent to the starfish's search for food sources, and realizes a detailed dimension-by-dimensional search of the parameters. Equation (8) is as follows:

[0088]

[0089] Where T is the current iteration number, A1 and A2 are two random numbers between (-1, 1). Formula (8) determines the new exploration direction by combining the current individual position, the relative position difference between the two random individuals, and the energy level of the starfish in a weighted summation method. p is a random number in D dimension, E t is the energy of the starfish, Et and θ can be obtained from equations (9) and (10):

[0090]

[0091] Among them, T max represents the maximum number of iterations. The parameter θ changes dynamically with the number of iterations (θ∈[0,π / 2]).

[0092] In the development phase, the algorithm first calculates the distance dm between the global optimal solution and other starfish individuals by formula (11) to determine the degree of closeness of each individual to the current optimal solution. Formula (11) is as follows:

[0093]

[0094] Among them, d m Represents the five calculated distances between the global optimal solution and other starfish individuals, m p Five randomly selected starfish individuals.

[0095] Then, for all starfish individuals except the one with the worst fitness, the system can update their positions through a parallel bidirectional search strategy and use Equation (12) to perform predation behavior, that is, update the parameter configuration according to the distance parameter dm to encourage the individual to move towards a better solution. Equation (12) is as follows:

[0096]

[0097] Among them, r1 and r2 are random numbers in the interval (0,1), d m1 and d m2 It is from d m Two distance parameters are randomly selected from [ ]. Based on a parallel bidirectional search strategy, some starfish candidate solutions move toward the optimal solution, while other candidate solutions move in the opposite direction within the same iteration cycle. This mechanism ensures that all candidate solutions have equal ability to break through the local optimal trap.

[0098] In addition, for the individual with the worst fitness (i.e., the starfish at the bottom of the ranking), the system can implement a regeneration strategy through equations (13) and (14), that is, significantly change its parameter configuration to break the state that may be trapped in the local optimal state and improve the search diversity and global convergence ability of the entire group. Equations (13) and (14) are as follows:

[0099]

[0100] After each round of iteration, the system will re-evaluate the fitness values ​​of all individuals and find the current local optimal solution through comparison, that is, the fuzzy neural network model with parameter configuration closest to the actual water pipe scaling thickness prediction.

[0101] Finally, the system updates the fitness value F and takes The smallest value As And check whether the maximum number of iterations Tmax is reached. If not (T <T max ), then return to the stage selection mechanism and continue the exploration or development of the next cycle. If the maximum number of iterations (T = T max ), then the best individual found in the entire cycle is output, that is, the most optimized parameter configuration End the optimization process.

[0102] Through the above process, the Starfish Optimization Algorithm plays a key role in parameter optimization during fuzzy neural network training. The algorithm alternates between exploration and exploitation phases, effectively optimizing the fuzzy neural network parameters through a combination of the direction angle θ, the energy variable Et, and a predation and regeneration strategy. This natural phenomenon-based optimization strategy effectively avoids local optimality and improves the performance and stability of the fuzzy neural network in predicting scale thickness in computer room water pipes.

[0103] In an optional embodiment, the development stage includes a parallel two-way search strategy and a regeneration strategy, wherein: the parallel two-way search strategy is used to use the distance parameter between two random individuals in the population as the update basis to control some candidate solutions of the population to move in a better direction, while other candidate solutions in the population except for this part of the candidate solutions perform reverse movement within the same iteration cycle; the regeneration strategy is used to update the position of the individual with the smallest fitness value in the population by moving at a preset speed to avoid the individual being determined as the local optimal solution.

[0104] Optionally, in the starfish optimization algorithm, the development phase searches for the global optimal solution by simulating the predation and regeneration behaviors of starfish, so two update strategies, namely a parallel bidirectional search strategy and a regeneration strategy, are designed in this phase.

[0105] Alternatively, to model the starfish predation behavior, the starfish optimization algorithm uses a parallel bidirectional search strategy that utilizes information about other starfish in the population and the current optimal position. The algorithm first calculates five sets of distance parameters between the optimal position and other starfish. Then, based on the parallel bidirectional search strategy, it randomly selects two distances as the update basis to update the position of each starfish. The distance parameters are shown in Formula (11).

[0106] In formula (11), d m Represents the five calculated distances between the global optimal solution and other starfish individuals, m p Five randomly selected starfish individuals.

[0107] Furthermore, due to their slow movement, starfish are vulnerable to predators while feeding. If captured by a predator, they can escape by self-severing their tentacles. Therefore, the starfish optimization algorithm's regeneration strategy only applies to the lowest ranked individual in the population (i = N).

[0108] Optionally, the above process significantly improves the accuracy and robustness of water pipe scale thickness prediction by integrating the development phase strategy of the starfish optimization algorithm with the fuzzy neural network model. The parallel bidirectional search strategy adopted in the development phase effectively avoids the dilemma of falling into the local optimal solution. By calculating the distance between the starfish individuals in the population and the global optimal position, the update basis is randomly selected to guide the parameters to evolve in a more optimized direction, thereby accelerating the convergence of the model. The regeneration strategy specifically targets the worst-performing individuals and increases the population diversity by significantly changing their parameter positions, ensuring the stability of the model prediction even under unfavorable conditions. The dynamic combination of these two strategies enables the fuzzy neural network model to not only efficiently explore the global solution space, but also perform in-depth optimization after identifying high-quality solutions, ultimately achieving highly accurate predictions of scale thickness in the water pipes of the computer room cooling system, providing a powerful technical tool for data center maintenance and decision-making.

[0109] In an optional embodiment, a one-dimensional search mode is used in the exploration phase to update the position of each individual in the population, including: detecting the position information of the i-th individual in the population at the t-th iteration, where the i-th individual is any individual in the population and t is an integer greater than or equal to 1; detecting the position information of any two randomly selected individuals in the population; and updating the position information of the i-th individual based on the position information of the i-th individual in the population at the t-th iteration, the position information of any two randomly selected individuals in the population, and a preset random number.

[0110] Alternatively, during the exploration phase of the starfish optimization algorithm, the system can utilize a novel search mode that combines a five-dimensional search mode (for dimensions D ≥ 5) with a one-dimensional search mode (for dimensions D < 5) for different optimization problems. Because the dimensionality of the optimization problem is less than 5 (D ≤ 5), the exploration phase uses a one-dimensional search mode to update the position. In this case, the starfish only moves one arm to search for food sources, utilizing the position information of other starfish. This is shown in the above formula (8).

[0111] In formula (8), T is the current iteration number, A1 and A2 are two random numbers between (-1, 1). This formula determines the new exploration direction by combining the current individual's position, the relative position difference between the two random individuals, and the energy level of the starfish in a weighted summation method. p is a random number in D dimension, E t is the energy of the starfish, which reflects the influence of the remaining number of iterations on the exploration intensity, as shown in equations (9) and (10).

[0112] Among them, T max The parameter θ changes dynamically with the number of iterations (θ∈[0,π / 2]), which determines the subtle changes in the exploration direction and helps to perform more refined searches in the later stages of the iteration.

[0113] For example, the position information of the i-th individual at the t-th iteration can be expressed as Indicates that the position information of any two randomly selected individuals can be expressed as A1 and A2 are two random numbers between (-1, 1), p is a random number in D dimension, and the position information of the i-th individual can be updated as

[0114] Optionally, as the number of iterations T increases, E t The value of is monotonically decreasing, reflecting the gradual consumption of starfish energy. t The value allows the starfish to explore a wider range; in the later stages of the iteration, when the energy is lower, it tends to conduct smaller local searches in order to find a more accurate position near the optimal solution.

[0115] Optionally, in one-dimensional search mode, the starfish optimization algorithm fine-tunes the parameters in the position update formula, including the energy variable E t , random factors A1 and A2, and the dimensionality factor p, enable efficient exploration of fuzzy neural network parameters in optimization problems with D ≤ 5. This exploration model not only reduces algorithmic complexity but also effectively avoids local optimality by dynamically adjusting the search scope and direction, providing a solid foundation for optimizing the water pipe scale thickness prediction model. In practical applications, this mechanism enables the model to quickly adapt to changes in water quality parameters and accurately predict future water pipe scaling conditions, thereby assisting in developing effective maintenance strategies for computer room cooling systems and ensuring the efficient and stable operation of bank data centers.

[0116] In an optional embodiment, the development phase searches for the global optimal solution by simulating the predation behavior of starfish, including: updating the position information of each individual in the population according to the position information of the i-th individual in the population at the t-th iteration, the position information of the j-th individual at the t-th iteration, and the current global optimal solution of the population at the t-th iteration, wherein the i-th individual is any individual in the population, and the j-th individual is an individual randomly selected from the population except the i-th individual; determining the global optimal solution of the population after the t-th iteration according to the updated position information of each individual in the population at the t-th iteration.

[0117] Alternatively, the update rule for each individual starfish in the predation behavior can be modeled as the above formula (12).

[0118] In formula (12), r1 and r2 are random numbers in the interval (0,1), d m1 and d m2 It is from d m Two distance parameters are randomly selected from [ ]. Based on a parallel bidirectional search strategy, some starfish candidate solutions move toward the optimal solution, while other candidate solutions move in the opposite direction within the same iteration cycle. This mechanism ensures that all candidate solutions have equal ability to break through the local optimal trap.

[0119] For example, the position information of the i-th individual at the t-th iteration can be expressed as The position information of the jth individual at the tth iteration can be expressed as The current global optimal solution can be expressed as Then the position information of the i-th individual can be updated to Y i T .

[0120] Optionally, during the predation phase of the starfish optimization algorithm, the update rule is updated by introducing random factors r1 and r2, combined with the distance parameter d from the global optimal solution to other individuals. m1 and d m2 , achieving parallel bidirectional search. This mechanism promotes population diversity, ensuring that all candidate solutions are capable of escaping local optima, and significantly enhances the global optimization capability and prediction accuracy of the fuzzy neural network model in the task of predicting water pipe scale thickness.

[0121] In an optional embodiment, the development phase searches for the global optimal solution by simulating the regeneration behavior of starfish, including: determining the movement speed of the i-th individual in the population at the t+1-th iteration based on the position information of the i-th individual in the population at the t-th iteration, the position information of the individual with the smallest fitness value in the population at the t-th iteration, and a preset minimum value; and determining the global optimal solution of the population at the t+1-th iteration based on the movement speed of each individual in the population at the t+1-th iteration.

[0122] Optionally, given that the biological regeneration process takes several months to complete (corresponding to the algorithm parameter adjustment), the starfish's movement speed during this stage is set to an extremely low value. Based on this, the position update rule for the regeneration stage is constructed as shown in the above formulas (13) and (14).

[0123] Although the number of function evaluations of the regeneration strategy is minimal (only a single calculation is performed during each iterative development phase), this phase is necessary to avoid local solutions and enhance global convergence.

[0124] For example, the position information of the i-th individual at the t-th iteration can be expressed as The individual with the smallest fitness value in the population at the tth iteration is the last individual in the population (i=N). The moving speed of the i-th individual at the t+1th iteration can be determined as

[0125] Optionally, in the regeneration phase, the starfish optimization algorithm uses an extremely slow moving speed update strategy to ensure that the weakest individual in the population has the opportunity to make large position adjustments, and can effectively explore the space far away from the current solution even in the later stage of the algorithm, thereby improving the global optimization ability of the fuzzy neural network in the water pipe scaling prediction problem and ensuring the high quality and stability of the prediction model.

[0126] Optionally, Figure 4 This is an overall flow chart of an optional method for determining the thickness of water pipe scale according to an embodiment of the present application, such as Figure 4As shown in Figure 1, the system first collects water quality data from the computer room cooling system, including water hardness, total dissolved solids (TDS), pH (acidity and alkalinity), flow rate, water temperature, and the time of data collection. This multi-dimensional information is integrated into an input vector for the fuzzy neural network model.

[0127] Next, the fuzzy network model creates an initial population of individual starfish. Each individual represents a set of parameter configurations for the fuzzy neural network, including the center value and width of the membership function and the weights of the neurons in the rule layer. These parameters form the basis for the model to predict the thickness of scale deposits in water pipes.

[0128] The system then uses the Starfish optimization algorithm to evaluate the model performance of each individual on the training dataset, specifically calculating its root mean square error (RMSE). This value serves as the individual's fitness function value and is used to measure the accuracy of the model's predictions.

[0129] During the iteration of model training, a random number is compared with the global stage selection threshold (default value is 0.5) to determine whether the current iteration will execute the exploration phase or the development phase. This mechanism ensures the algorithm maintains a balance between global optimization and local refinement.

[0130] When the random number exceeds the stage selection threshold, the system enters the exploration phase. Starfish individuals adjust their parameters according to the starfish optimization algorithm's strategy, performing a single-dimensional search of the parameter space by calculating azimuth and energy variables. The goal is to discover a wider range of optimal solutions and avoid falling into local optima.

[0131] When the random number is less than the stage selection threshold, the system enters the development stage. The development stage focuses on optimizing the positions of individuals in the current population, updating the positions of most individuals through predation strategies, and adjusting the parameters of individuals with the worst fitness using regeneration strategies to improve the accuracy of model predictions and global convergence capabilities.

[0132] After each iteration, the system will update the positions of all individuals and recalculate the fitness value to determine the current local optimal solution. In addition, the system will monitor the number of iterations and stop when the preset maximum number of iterations is reached.

[0133] When the maximum number of iterations is reached or the fitness value converges to meet the conditions, the system will output the individual parameter configuration with the lowest fitness value in the entire optimization process, that is, the optimal parameter configuration of the fuzzy neural network.

[0134] Finally, the optimal parameters determined by the Starfish optimization algorithm were imported into the trained fuzzy neural network. The collected water quality data, including water hardness, TDS, pH, flow rate, temperature, and time, were fed into the network as input vectors. After Gaussian distribution transformation in the fuzzification layer, fitness calculation in the rule layer, standardization in the normalization layer, and comprehensive evaluation in the output layer, a clear prediction result—the predicted value of scale thickness in the water pipe—was produced.

[0135] After completing the entire prediction process, the system outputs the predicted results of the scale thickness of the computer room water pipes at a specified time point, providing a scientific decision-making basis for the maintenance and management of the computer room cooling system.

[0136] In summary, Figure 4 The method presented here effectively improves the accuracy of fuzzy neural network predictions for scale thickness in computer room water pipes using the Starfish optimization algorithm. The algorithm's dynamic switching between the exploration and development phases ensures the model can comprehensively search the parameter space while fine-tuning parameters, avoiding local optimality and significantly enhancing global convergence. Ultimately, the optimized fuzzy neural network model can more accurately analyze water quality data and efficiently predict water pipe scaling, providing strong support for cooling system maintenance.

[0137] According to another aspect of the present application, a device for determining the thickness of scale in a water pipe is provided, wherein: Figure 5 is a schematic diagram of an optional device for determining the thickness of water pipe scale according to an embodiment of the present application, such as Figure 5 As shown, the device for determining the thickness of scale in a water pipe includes: a collecting unit 501 and a determining unit 502 .

[0138] Optionally, the collection unit 501 is used to use the water hardness information, total dissolved solids information, pH information, water flow rate information, water temperature information and collection time information collected from the water pipe as model input variables; the determination unit 502 determines the scale thickness of the water pipe at the target time based on the model input variables through the prior knowledge learned by the fuzzy neural network model in the model training stage, wherein the training objectives of the fuzzy neural network model include: adjusting the membership and weight parameters of the fuzzy neural network using the starfish optimization algorithm according to the training sample set and the network parameter set, until the loss function value of the fuzzy neural network reaches a preset threshold.

[0139] Optionally, the apparatus for determining the thickness of scale deposits on a water pipe further includes: a first processing unit and a second processing unit. The first processing unit is configured to initialize a network parameter set of a fuzzy neural network, wherein the network parameter set includes at least the center value of a membership function, the width of the membership function, and a weight value; and the second processing unit is configured to adjust the membership and weight parameters of the fuzzy neural network using a starfish optimization algorithm based on the successfully initialized network parameter set and a training sample set until the loss function value of the fuzzy neural network reaches a preset threshold, thereby training a fuzzy neural network model.

[0140] Optionally, the determination unit 502 includes: a first detection subunit, a second detection subunit, and a first updating subunit. The first detection subunit is configured to detect the location information of the i-th individual in the population at the t-th iteration, where the i-th individual is any individual in the population and t is an integer greater than or equal to 1; the second detection subunit is configured to detect the location information of any two randomly selected individuals in the population; and the first updating subunit is configured to update the location information of the i-th individual based on the location information of the i-th individual in the population at the t-th iteration, the location information of any two randomly selected individuals in the population, and a preset random number.

[0141] Optionally, the determination unit 502 further includes: a second updating subunit and a first determining subunit. The second updating subunit is configured to update the position information of each individual in the population based on the position information of the i-th individual in the population at the t-th iteration, the position information of the j-th individual at the t-th iteration, and the current global optimal solution of the population at the t-th iteration, wherein the i-th individual is any individual in the population, and the j-th individual is an individual randomly selected from the population except the i-th individual; and the first determining subunit is configured to determine the global optimal solution of the population after the t-th iteration based on the updated position information of each individual in the population at the t-th iteration.

[0142] Optionally, the determining unit 502 further includes: a second determining subunit and a third determining subunit. The second determining subunit is configured to determine the movement speed of the i-th individual in the population at the t+1th iteration based on the position information of the i-th individual in the population at the tth iteration, the position information of the individual with the smallest fitness value in the population at the tth iteration, and a preset minimum value; and the third determining subunit is configured to determine the global optimal solution of the population at the t+1th iteration based on the movement speed of each individual in the population at the t+1th iteration.

[0143] According to another aspect of the present application, a computer-readable storage medium is provided, comprising: a computer program stored in the computer-readable storage medium, wherein when the computer program is executed, the device where the computer-readable storage medium is located executes the above-mentioned method for determining the thickness of scale in water pipes.

[0144] According to another aspect of the present application, an electronic device is also provided, comprising one or more processors and a memory, wherein the memory is used to store one or more programs, wherein when the one or more programs are executed by one or more processors, the one or more processors execute the above-mentioned method for determining the thickness of scale in a water pipe.

[0145] The above-mentioned embodiments or examples disclosed in this application are not exhaustive, but are only illustrations of some embodiments or examples, and are not intended to be specific limitations on the scope of protection disclosed in this application. In the absence of contradiction, each step in a certain embodiment or example in this application can be implemented as an independent example, and the steps can be arbitrarily combined. For example, the solution after removing some steps in a certain embodiment or example can also be implemented as an independent example, and the order of the steps in a certain embodiment or example can be arbitrarily exchanged. In addition, the optional methods or optional examples in a certain embodiment or example can be arbitrarily combined; in addition, the various embodiments or examples can be arbitrarily combined. For example, some or all of the steps in different embodiments or examples can be arbitrarily combined, and a certain embodiment or example can be arbitrarily combined with the optional methods or optional examples of other embodiments or examples.

[0146] The serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0147] In the above embodiments of the present application, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, please refer to the relevant description of other embodiments.

[0148] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only exemplary. For example, the division of units can be a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0149] Units described as separate components may or may not be physically separate, and components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple units. Some or all of the units may be selected to achieve the purpose of the present embodiment according to actual needs.

[0150] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0151] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for enabling a computer device (which can be a personal computer, server or network device, etc.) to execute all or part of the steps of the various embodiments of the present application. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk and other media that can store program code.

[0152] The above is only a preferred embodiment of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A method for determining the thickness of scale buildup in water pipes in a data center computer room, characterized in that: include: The water hardness information, total dissolved solids information, pH value information, water flow rate information, water temperature information and collection time information collected from the water pipe are used as model input variables; The prior knowledge learned by the fuzzy neural network model during the model training phase is used to determine the scale thickness of the water pipe at the target time based on the model input variables. The training objectives of the fuzzy neural network model include: adjusting the membership and weight parameters of the fuzzy neural network using the starfish optimization algorithm based on the training sample set and the network parameter set until the loss function value of the fuzzy neural network reaches a preset threshold.

2. The method according to claim 1, characterized in that The training process of the fuzzy neural network model includes the following steps: Initializing a network parameter set of the fuzzy neural network, wherein the network parameter set includes at least: a central value of a membership function, a width of the membership function, and a weight; According to the network parameter set after successful initialization and the training sample set, the membership and weight parameters of the fuzzy neural network are adjusted using the starfish optimization algorithm until the loss function value of the fuzzy neural network reaches the preset threshold, and the fuzzy neural network model is trained.

3. The method according to claim 1, characterized in that The fuzzy neural network model includes an input layer, a fuzzification layer, a rule layer, a normalization layer and an output layer, wherein: The input layer is used to input the model input variables into the fuzzy neural network model, wherein the number of nodes in the input layer is equal to the number of dimensions of the model input variables; The fuzzification layer is used to perform fuzzification processing on the model input variables and map the input space to a Gaussian distribution space, wherein the number of nodes in the fuzzification layer is equal to the number of fuzzy levels of the model neural network model; each node in the fuzzification layer represents a fuzzy speech variable value and is used to calculate the membership function of the components of the input vector to the fuzzy set of the respective speech variables; The rule layer is used to calculate the total fuzzy membership of the model input variables through fuzzy rules, wherein the number of summary points of the rule layer is equal to the number of fuzzy rules, and each node of the rule layer is used to determine the applicability of a fuzzy rule; The normalization layer is used to perform normalization calculation on the applicability of each fuzzy rule, wherein the total number of nodes in the normalization layer is equal to the total number of nodes in the rule layer; The output layer is used to integrate the fuzzy rules to determine the output result of the fuzzy neural network model.

4. The method according to claim 1, wherein The following stages are followed when executing the Starfish optimization algorithm: The initialization phase is used to generate a position initialization matrix, evaluate the fitness values ​​of all individuals in the population through the objective function, and construct a matrix for storing and updating the fitness values ​​based on the position initialization matrix; The selection phase is used to select whether to enter the exploration phase or the development phase based on the comparison result of the random number and the global phase selection threshold; The exploration phase is used to update the position of each individual in the population using a one-dimensional search mode, wherein the one-dimensional search mode is used to update the position of the current individual by simulating the exploration behavior of starfish tentacles and using the position information of other individuals in the population except the current individual; The development phase is used to search for a global optimal solution by simulating the predation or regeneration behavior of starfish; The iterative cycle stage is used to update the fitness values ​​of all individuals in the population after each iteration and store the current optimal solution; if the population does not meet the termination condition after any iteration, it returns to the selection stage to start a new cycle; if the population meets the termination condition after any iteration, the optimal solution obtained in the last round is retained as the final output result, wherein the final output result is used to optimize the parameters of the fuzzy neural network.

5. The method according to claim 4, characterized in that The development phase includes a parallel bidirectional search strategy and a regeneration strategy, wherein: The parallel bidirectional search strategy is used to use the distance parameter between two random individuals in the population as an update basis to control some candidate solutions of the population to move towards a better direction, while other candidate solutions in the population except the candidate solutions perform reverse movement within the same iteration cycle; The regeneration strategy is used to update the position of the individual with the smallest fitness value in the population by moving the individual at a preset speed, so as to prevent the individual from being determined as a local optimal solution.

6. The method according to claim 4, characterized in that The exploration phase uses a one-dimensional search mode to update the position of each individual in the population, including: Detecting the position information of the i-th individual in the population at the t-th iteration, wherein the i-th individual is any individual in the population, and t is an integer greater than or equal to 1; Detecting position information of any two randomly selected individuals in the population; The position information of the i-th individual in the population is updated according to the position information of the i-th individual in the population at the t-th iteration, the position information of any two randomly selected individuals in the population, and a preset random number.

7. The method according to claim 4, characterized in that The development phase searches for the global optimal solution by simulating the starfish's predation behavior, including: Update the position information of each individual in the population according to the position information of the i-th individual in the population at the t-th iteration, the position information of the j-th individual at the t-th iteration, and the current global optimal solution of the population at the t-th iteration, wherein the i-th individual is any individual in the population, and the j-th individual is an individual randomly selected from the population except the i-th individual; According to the updated position information of each individual in the population at the t-th iteration, a global optimal solution of the population after the t-th iteration is determined.

8. The method according to claim 4, characterized in that The development phase searches for the global optimal solution by simulating the regeneration behavior of starfish, including: Determine the movement speed of the i-th individual at the t+1-th iteration according to the position information of the i-th individual in the population at the t-th iteration, the position information of the individual with the smallest fitness value in the population at the t-th iteration, and a preset minimum value; According to the movement speed of each individual in the population at the t+1th iteration, a global optimal solution of the population at the t+1th iteration is determined.

9. A device for determining the thickness of scale buildup in water pipes in a data center computer room, characterized in that: include: A collection unit is used to use water hardness information, total dissolved solids information, pH value information, water flow rate information, water temperature information and collection time information collected from the water pipe as model input variables; A determination unit determines the scale thickness of the water pipe at a target time based on the prior knowledge learned by the fuzzy neural network model during the model training phase and the model input variables, wherein the training objectives of the fuzzy neural network model include: adjusting the membership and weight parameters of the fuzzy neural network using the starfish optimization algorithm according to the training sample set and the network parameter set until the loss function value of the fuzzy neural network reaches a preset threshold.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located executes the method for determining the scale thickness of water pipes based on a data center computer room according to any one of claims 1 to 8.

11. An electronic device, characterized in that: The method comprises one or more processors and a memory, wherein the memory is used to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors execute the method for determining the scale thickness of water pipes in a data center computer room according to any one of claims 1 to 8.

12. A computer program product, characterized in that The method comprises a computer program or an instruction, which, when executed by a processor, implements the method for determining the scale thickness of water pipes in a data center computer room according to any one of claims 1 to 8.

Images