An equipment loading system fault diagnosis method

Abstract

The application relates to the technical field of equipment loading system, and discloses an equipment loading system fault diagnosis method, which collects fault characteristic parameters of an equipment loading system as original data; a tuna algorithm TSO is improved, including introducing a Logistic chaotic mapping in the TSO initialization stage, introducing a nonlinear adjustment strategy in the TSO spiral hunting stage to enhance the global search ability and the local development ability of the algorithm, and introducing a mutation method in the differential evolution method to avoid the singleness of the population, using the improved tuna algorithm ITSO to optimize the weight of a TOPSIS algorithm, constructing an ITSO-TOPSIS data processing model, optimizing parameters of a long short-term memory network LSTM algorithm, and constructing an ITSO-LSTM fault diagnosis model, etc. The application improves the performance of the two algorithms, can be more efficient and accurate in the fault diagnosis process, provides technical support for the maintenance of the equipment loading system, and remedies the defects of blindness in parameter selection in the training process.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of equipment loading system, in particular to a kind of equipment loading system fault diagnosis method. BACKGROUND

[0002] Automatic loading system as an important part of equipment, its performance and work reliability have key role to the promotion of the overall combat capability of equipment.However, automatic loading system is a control logic highly complex and prone to mechanical failure system, once fault occurs in the running process, fault identification and positioning are often very difficult, which may cause serious impact on the combat capability of equipment.As a complex system with time sequence control and action control characteristics, its multi-level logic relationship and strict timing requirements make comprehensive and efficient fault diagnosis become a challenging research topic.Therefore, how to improve the fault diagnosis capability of automatic loading system becomes one of the core problems to guarantee the combat performance of equipment.

[0003] Traditional fault diagnosis method is too dependent on manpower, resource consumption is large and diagnosis efficiency is low, which is difficult to meet the needs of complex system.At the same time, the fault diagnosis rate of these methods in complex system is relatively limited.In recent years, fault diagnosis technology based on intelligent algorithm has gradually become an important research direction in the field of complex system fault diagnosis.In the automatic loading system of equipment, the component composition is complex, and the circuit logic is highly coupled, so accurate component level positioning of the system is needed in the fault diagnosis process, and specific maintenance suggestions are given, so as to realize comprehensive state monitoring and accurate fault diagnosis of automatic loading system. SUMMARY

[0004] In view of the above shortcomings and deficiencies in the prior art, the present application provides a kind of equipment loading system fault diagnosis method, improves tuna algorithm TSO, the weight of TOPSIS algorithm is optimized by improved tuna algorithm ITSO, the key parameters of LSTM algorithm are optimized, the fault diagnosis model is constructed, the defect of blindness in parameter selection in training process is made up, including the following steps:

[0005] Step S01, the fault characteristic parameters of equipment loading system are collected as original data;

[0006] Step S02, improve tuna algorithm TSO, including introducing Logistic chaotic mapping in TSO initialization stage, introducing nonlinear adjustment strategy in TSO spiral hunting stage to enhance the global search ability and local development ability of algorithm, and introducing mutation method in differential evolution method to avoid the singleness of population, to obtain improved tuna algorithm ITSO;

[0007] Step S03, the improved tuna algorithm ITSO is used to optimize the weight of TOPSIS algorithm, an ITSO-TOPSIS data processing model is constructed, the original data collected in step S01 is sorted by correlation degree information through the ITSO-TOPSIS data processing model, the key attribute information affecting the equipment loading system is screened out, a data set is constructed, and the data set is divided into a training set and a test set;

[0008] Step S04, the improved tuna algorithm ITSO is used to optimize the parameters of the long short-term memory network LSTM algorithm, and an ITSO-LSTM fault diagnosis model is constructed;

[0009] Step S05, the training set in step S03 is used to train the ITSO-LSTM fault diagnosis model in step S04;

[0010] Step S06, the test set in step S03 is used to test the ITSO-LSTM fault diagnosis model trained in step S05;

[0011] Step S07, the ITSO-LSTM fault diagnosis model that passes the test is used to predict the health of the equipment loading system, and the prediction result is output.

[0012] In step S02, the Logistic chaotic mapping is introduced in the TSO initialization stage, and the improved formula is:

[0013] ;

[0014] In the formula: is the initial position of the ith individual after Logistic mapping; X i is the ith individual randomly generated before Logistic mapping; ub is the upper boundary of the search space; lb is the lower boundary of the search space; N is the number of tuna swarm, and R is the chaotic control parameter;

[0015] The parameter a that determines the degree of following the best individual of the previous individual in the initial stage is set, a nonlinear adjustment strategy is introduced to dynamically adjust a, and the formula is:

[0016] ;

[0017] Wherein; a max and a min are the initial value and the minimum value respectively; is a nonlinear control factor;

[0018] The improved a is input into the original mathematical model, and the weight coefficient alpha 1 of the trend of moving the control individual to the previous individual and the weight coefficient alpha 2 of the trend of moving the control individual to the optimal individual are updated:

[0019] ;

[0020] ;

[0021] Wherein, t is the current iteration number, t max is the maximum iteration number;

[0022] The spiral search reference point used in the spiral foraging stage will gradually change from a randomly selected individual to the updated position of the current optimal individual, and the mathematical model is as follows:

[0023] ;

[0024] Wherein, is the individual of the t+1 iteration, is a randomly generated spiral search reference point in the search space, is an intermediate variable, is the current optimal individual, is the individual of the t iteration, and rand is a random number in the range [0, 1];

[0025] At this time, the mutation method in the differential evolution method is introduced to realize individual mutation to generate a mutated solution, and then the mutated solution and the original solution are combined to generate a crossover solution, thereby avoiding the singleness of the population and judging whether the new individual is adopted according to the value of the fitness function, and the calculation formula is as follows:

[0026] ;

[0027] ;

[0028] ;

[0029] Wherein: is the position of the mutated individual; both represent an individual randomly selected in the population; is the position of the individual after crossover; F is a scaling factor with a value range of [0.4, 1]; and CR is a crossover probability factor with a value range of [0, 2].

[0030] Step S04 performs parameter optimization on the long short-term memory network (LSTM) algorithm by using the improved tuna optimization (ITSO) algorithm, and the optimized parameters include a weight factor, a learning rate and a bias term.

[0031] Compared with the prior art, the application has the following beneficial technical effects and advantages:

[0032] The application improves the tuna optimization algorithm TSO by introducing the Logistic chaotic mapping, the nonlinear adjustment strategy and the mutation method in the differential evolution method, enhances the global search ability, the precision and the efficiency in local development, avoids the singleness of the population, and has higher applicability to different algorithm parameters.

[0033] The improved tuna optimization algorithm is used for optimizing the key parameters of the TOPSIS and the LSTM, improves the performance of the two algorithms, can be more efficient and accurate in the process of fault diagnosis, and provides technical support for the maintenance of the equipment loading system. BRIEF DESCRIPTION OF DRAWINGS

[0034] Figure 1 The application provides an equipment loading system fault diagnosis method flow chart. DETAILED DESCRIPTION

[0035] The application will be described in detail below, and the protection scope of the application is not limited by the drawings.

[0036] The application provides an equipment loading system fault diagnosis method, which comprises the following steps:

[0037] Step S01, collecting fault characteristic parameters of the equipment loading system as original data.

[0038] Step S02, improving the tuna optimization algorithm TSO, including introducing the Logistic chaotic mapping in the TSO initialization stage, introducing the nonlinear adjustment strategy in the TSO spiral hunting stage to enhance the global search ability and the local development ability of the algorithm, and introducing the mutation method in the differential evolution method to avoid the singleness of the population, to obtain the improved tuna optimization algorithm ITSO.

[0039] Specifically,

[0040] 1) initialization stage: introducing the Logistic chaotic mapping, and the improved formula is:

[0041] ;

[0042] In the formula, is the initial position of the ith individual after the Logistic mapping; X i is the ith individual randomly generated before the Logistic mapping; ub is the upper boundary of the search space; lb is the lower boundary of the search space; N is the number of tuna groups, and R is the chaotic control parameter.

[0043] 2) Spiral hunting phase: based on the behavior of spiral hunting of tuna school in nature, and combined with the characteristics of information transmission between tuna, the original mathematical model is constructed as follows:

[0044] ; ; ; ; ;

[0045] In the formula: is the current best individual, is the individual of t+1 iteration, is the individual of t iteration, is the weight coefficient of the trend of the individual moving to the previous individual, is the weight coefficient of the trend of the individual moving to the best individual, a is a constant that determines the degree of following the previous individual and the best individual in the initial stage of the individual, is a random number in (0, 1], and l are intermediate variables; t max is the maximum number of iterations.

[0046] To avoid long-time search without finding prey due to blindly following the best individual. Therefore, the existing best individual can be discarded, and a coordinate randomly selected in the entire search space can be taken as a new best individual, which is beneficial to foraging hunting in the entire space. The corresponding mathematical model is as follows: ;

[0047] In the formula: is a randomly generated point in the search space as a spiral search reference.

[0048] Since the tuna algorithm adjusts global search and local development according to the value of a, a is linearly changed in the original algorithm, which is difficult to adapt to different optimization problems. Therefore, a nonlinear adjustment strategy is introduced here to dynamically adjust a according to the current iteration number, so as to more flexibly control the balance between global search and local development. The new formula after improvement is:

[0049] ;

[0050] Wherein; a max and a min are the initial value and the minimum value respectively; is a nonlinear control factor;

[0051] The improved a is brought into the original mathematical model to update the weight coefficient α1 of the trend of the individual moving to the previous individual and the weight coefficient α2 of the trend of the individual moving to the best individual, so as to enhance the global search ability and local development ability of the algorithm.

[0052] The tuna school algorithm starts with a wide range of global search, and then gradually narrows down the range, one area after another. With the gradual increase of the number of iterations, the reference point used in the spiral foraging stage will gradually change from randomly selected individuals to the current optimal individual. The mathematical model is as follows: ;

[0053] Wherein, is the individual of t+1 iteration, is randomly generated in the search space as a spiral search reference point, is the current best individual, is the individual of t iteration, and rand is a random number in the range [0, 1].

[0054] After the above position update, the fitness value of the current position is initialized and compared with the fitness value of the previous position. The optimized offspring is put into the next iteration. This method of updating the individual position without interference is prone to the problem of local optimum with the increase of the number of iterations. Therefore, the mutation method in differential evolution is introduced to realize individual mutation to produce mutation solution, and then the crossover solution is generated by combining the mutation solution with the original solution, which avoids the single population and judges whether to use the new individual according to the value of the fitness function. The calculation formula is as follows:

[0055] ;

[0056] ;

[0057] ;

[0058] Wherein: is the position of the individual after mutation; both represent the individuals randomly selected in the population; is the position of the individual after crossover; F is a scaling factor with a value range of [0.4, 1]; CR is a crossover probability factor with a value range of [0, 2].

[0059] 3) Parabolic hunting stage: when the tuna school is arranged in a parabolic shape for hunting, it will also search for prey around it. The probability of selecting these two methods is set to 50% each. The mathematical model is as follows:

[0060] ;

[0061] ;

[0062] In the formula: TF is a random number with a value of 1 or -1, and p is an adjustment factor that changes with the number of iterations.

[0063] Step S03, the improved tuna algorithm ITSO is used to optimize the weight of TOPSIS algorithm, an ITSO-TOPSIS data processing model is constructed, the original data collected in step S01 are sorted by correlation degree information through the ITSO-TOPSIS data processing model, the key attribute information affecting the equipment loading system is screened out, a data set is constructed, and the data set is divided into a training set and a test set.

[0064] Specifically, the TOPSIS algorithm steps are as follows:

[0065] Construct a decision matrix: let be an evaluation object with parameters, a * decision matrix is obtained.

[0066] Standardize the decision matrix: in order to eliminate the dimension influence, the decision matrix is standardized to obtain a standardized matrix:

[0067] ;

[0068] wherein is the standardized value, indicating the standardized result of the jth index of the ith sample; x ij is the original index value.

[0069] Construct a weighted standardized matrix: the improved tuna algorithm is introduced here to optimize the weight of each index, and the weighted standardized matrix is obtained by multiplying the optimized weight and the standardized matrix:

[0070] ;

[0071] wherein is the weight of the jth index; is the weighted index value.

[0072] Determine the ideal solution and the negative ideal solution: according to the weighted standardized matrix, the ideal solution (the best value) and the negative ideal solution (the worst value) of each index are determined:

[0073] ;

[0074] wherein: (if it is a benefit type index) (if it is a cost type index);

[0075] (if it is a benefit type index) (if it is a cost type index);

[0076] Calculate the distance of each sample to the ideal solution and the negative ideal solution: Calculate the Euclidean distance of the i th sample to the ideal solution and the negative ideal solution :

[0077] ;

[0078] Where: the distance of the i th sample to the ideal solution; the distance of the i th sample to the negative ideal solution.

[0079] Calculate the relative closeness: Calculate the relative closeness C of the i th sample with respect to the ideal solution i :

[0080] ;

[0081] Where: is the relative closeness of the i th sample.

[0082] According to the relative closeness C i , arrange the samples in descending order to obtain the final ranking of advantages and disadvantages.

[0083] Step S04, the improved tuna algorithm ITSO is used to optimize the parameters of the long short-term memory network LSTM algorithm, and the optimized parameters include weight factors, learning rates, bias terms, and an ITSO-LSTM fault diagnosis model is constructed.

[0084] Specifically, the steps of the long short-term memory network (LSTM) algorithm are as follows:

[0085] (1) Forget gate calculation

[0086] The forget gate is used to determine how much information from the cell state of the previous time is discarded.

[0087] The formula is as follows: ;

[0088] Where: is the output of the forget gate (the value range is 0 to 1), indicating the degree of forgetting of each cell state information; is the Sigmoid activation function, which ensures that the output value is in the range [0, 1]; is the weight matrix of the forget gate; is the spliced input, including the hidden state of the previous time and the current input x t ; Bias term for the forget gate.

[0089] (2) Input gate calculation

[0090] The input gate is used to decide how much new information is added to the cell state. It consists of two parts: the input gate weight t and the candidate state .

[0091] Input gate weight calculation: ;

[0092] where: is the output of the input gate (ranging from 0 to 1), representing the degree of influence of the current input on the cell state; is the weight matrix of the input gate; is the bias term of the input gate.

[0093] Candidate state calculation: ;

[0094] where: is the candidate cell state at the current time, representing the information that can be added to the cell state; is the hyperbolic tangent activation function, used to compress the candidate state value to [-1, 1]; is the weight matrix of the candidate state; is the bias term of the candidate state.

[0095] (3) Update cell state

[0096] Combine the results of the forget gate and input gate calculations to update the cell state C t at the current time: ;

[0097] where: is the cell state at the current time; is the cell state at the previous time; is the output of the forget gate, controlling the information of the cell state that needs to be retained; is the output of the input gate, controlling the addition of new information; is the current candidate state.

[0098] (4) Output gate calculation

[0099] The output gate decides the hidden state h t that needs to be output at the current time. It is calculated in two steps:

[0100] Output gate weight calculation: ;

[0101] where: Output (ranging from 0 to 1) of the output gate, indicating which information in the cell state needs to be output; Weight matrix of the output gate; Bias term of the output gate.

[0102] Hidden state update: ;

[0103] Wherein: Hidden state at the current moment, as the output at the current moment; Current cell state; Compress the value of the cell state to the range [-1, 1].

[0104] Step S05, using the training set in step S03 to train the ITSO-LSTM fault diagnosis model in step S04;

[0105] Step S06, using the test set in step S03 to test the ITSO-LSTM fault diagnosis model trained in step S05;

[0106] Step S07, using the test qualified ITSO-LSTM fault diagnosis model to predict the health of the equipment loading system, and output the prediction result.

[0107] The application improves the tuna optimization algorithm TSO by introducing the Logistic chaotic mapping, the nonlinear adjustment strategy and the mutation method in the differential evolution method, enhances the global search ability, the accuracy and the efficiency in local development, avoids the singleness of the population, and has higher applicability to different algorithm parameters.

[0108] The improved tuna optimization algorithm is used to optimize the TOPSIS algorithm and the LSTM key parameters, improve the performance of the two algorithms, and be more efficient and accurate in the process of fault diagnosis, and provide technical support for the maintenance of the equipment loading system.

[0109] Although the embodiments of the application have been shown and described above, it should be understood that the above embodiments are exemplary and should not be construed as limiting the application, and any modification, replacement, modification and variation of the above embodiments by those skilled in the art are within the scope of the application.

[0110] The above only describes the preferred embodiments of the application and should not be construed as limiting the application, any modification, equivalent replacement, improvement, etc. within the spirit and principle of the application should be included in the protection scope of the application.

Claims

1. A method for diagnosing faults in an equipment loading system, characterized in that, Includes the following steps: Step S01: Collect fault characteristic parameters of the equipment loading system as raw data; Step S02: Improve the tuna algorithm TSO by introducing Logistic chaotic mapping in the TSO initialization stage, introducing a nonlinear adjustment strategy in the TSO spiral hunting stage to enhance the algorithm's global search capability and local exploitation capability, and introducing the mutation method in differential evolution to avoid the homogeneity of the population, thus obtaining the improved tuna algorithm ITSO. Specifically, a Logistic chaotic mapping is introduced in the TSO initialization phase, and the improved formula is as follows: In the formula: X is the initial position of the i-th individual after the Logistic mapping; i is the i-th individual randomly generated before the Logistic mapping; ub is the upper boundary of the search space; lb is the lower boundary of the search space; N is the number of tuna; and R is the chaos control parameter. Let parameter 'a' be defined to determine the optimal degree to which an individual follows the previous individual in the initial stage. A nonlinear adjustment strategy is introduced to dynamically adjust 'a', as shown in the formula: in; a max and a min These are the initial value and the minimum value, respectively. δ is a nonlinear control factor; Substituting the improved 'a' into the original mathematical model, we update the weight coefficient α1 controlling the tendency of an individual to move towards the previous individual and the weight coefficient α2 controlling the tendency of an individual to move towards the optimal individual: Where t is the current iteration number, t max This represents the maximum number of iterations. During the spiral foraging phase, the spiral search reference point used will gradually change from a randomly selected individual to the current optimal individual's updated position. The mathematical model is as follows: in, For the individual in iteration t+1, The reference point for the spiral search is randomly generated in the search space, and β is an intermediate variable. As the best individual at present, For each individual in the t iterations, rand is a random number within the range [0, 1]. At this point, the mutation method from differential evolution is introduced to generate mutated solutions by individual mutation. These mutated solutions are then combined with the original solutions to generate crossover solutions, thus avoiding the homogeneity of the population. Based on the value of the fitness function, it is determined whether to adopt a new individual. The calculation formula is as follows: in: The location of the mutated individual; Both refer to individuals randomly selected from the population; is the orientation of the individual after crossover; F is a scaling factor with a value range of [0.4, 1]; CR is a crossover probability factor with a value range of [0, 2]. Step S03: The improved tuna algorithm ITSO is used to optimize the weights of the superior solution distance method TOPSIS algorithm, and an ITSO-TOPSIS data processing model is constructed. The original data collected in step S01 is sorted by correlation information through the ITSO-TOPSIS data processing model, and key attribute information that affects the equipment loading system is selected. A dataset is constructed and divided into training set and test set. Specifically, the TOPSIS algorithm steps include: Construct the decision matrix; Standardized decision matrix; Constructing a weighted standardized matrix: Here, the improved tuna algorithm is introduced to optimize the weights of each indicator. The standardized matrix is ​​multiplied by the optimized weights to obtain the weighted standardized matrix. Determine the ideal and negative ideal solutions: Based on the weighted standardization matrix, determine the ideal and negative ideal solutions for each index; Calculate the distance from each sample to the ideal solution and the negative ideal solution; Calculate the relative proximity; The samples are sorted in descending order based on their relative similarity to obtain the final ranking of their merits. Step S04: The improved tuna algorithm ITSO is used to optimize the parameters of the long short-term memory network LSTM algorithm and construct the ITSO-LSTM fault diagnosis model. Step S05: Train the ITSO-LSTM fault diagnosis model in step S04 using the training set from step S03. Step S06: Test the ITSO-LSTM fault diagnosis model trained in step S05 using the test set from step S03. Step S07: Use the tested and qualified ITSO-LSTM fault diagnosis model to perform health prediction on the equipment loading system and output the prediction results.

2. The fault diagnosis method for an equipment loading system according to claim 1, characterized in that: Step S04 uses the improved tuna algorithm ITSO to optimize the parameters of the long short-term memory network LSTM algorithm. The optimized parameters include weight factor, learning rate, and bias term.

Images