Method for adaptive neuro-fuzzy inference based energy management strategy for fuel cell ships

Abstract

The application discloses a kind of based on adaptive neural fuzzy inference fuel cell ship energy management strategy method, comprising the following steps: obtaining the topological structure of fuel cell ship hybrid power system, and its operating typical working condition information, the steps of obtaining the globally optimal hybrid power system load distribution by offline algorithm;By discrete solving working condition every moment optimal equivalent factor, the step of taking working condition calculation result as training sample;Build and train adaptive neural fuzzy inference system (ANFIS), calculate to obtain optimal equivalent factor, the step of bringing result into ECMS algorithm, on-line calculation ship sailing working condition obtains optimal distribution result, the improvement of the present application is realized on-line ideal power distribution by ANFIS calculation ECMS equivalent factor, combined with the calculation characteristics of three strategies, can on-line calculation obtain the optimal solution of ship real-time power distribution, with good global optimization ability.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of ship energy management strategy, in particular, the application method of power adaptability of fuel cell ship for high fluctuation load and fuel economy requirement under full voyage. BACKGROUND

[0002] In the field of vehicle fuel cell power system, due to the slow dynamic response of single fuel cell, it cannot quickly track the load change to meet the daily driving demand, so it must be equipped with a battery as a buffer unit or a range extender, thereby requiring its energy management strategy to control the target of power adaptability, fuel economy and endurance capability. Especially in the face of the characteristics of frequent start and stop and load change of vehicles, its energy management strategy prioritizes endurance capability. However, for ship fuel cell electric propulsion system, the endurance capability of the ship is also limited by the amount of hydrogen storage. But compared with vehicles, the ship has sufficient load and space, and the requirement for hydrogen storage system is not as strict as that of vehicles. Therefore, under the premise that the endurance capability of the ship is guaranteed, the main problem it faces is that there are many high-power equipment on the ship, and due to the influence of the route environment and the channel during the voyage, the working condition changes complexly, especially the load fluctuates at a high frequency during the start and stop process of the ship, which increases the difficulty of energy management system design. Therefore, the ship fuel cell energy management strategy focuses more on power adaptability for high fluctuation load and fuel economy under full voyage.

[0003] The common methods of ship fuel cell energy management strategy include rule-based strategy, global optimization-based strategy and instantaneous optimization-based strategy. Among them, the rule-based energy management strategy is too dependent on empirical values, and can only meet the power adaptability, but it is difficult to achieve precise control of fuel economy. Dynamic programming (DP) is a kind of offline algorithm based on global optimization, which can achieve global optimization, but it is too dependent on the computing power of the energy management system when facing multi-dimensional variables and complex equations, and it takes a long time, so it is difficult to be used for real-time control. Equivalent hydrogen consumption minimum algorithm (ECMS) is an online algorithm based on instantaneous optimization algorithm, its control performance depends on the selection of equivalent factor, and the optimization effect is worse than the global optimal solution of DP algorithm, which still has room for improvement. Therefore, for the power adaptability of fuel cell ship for high fluctuation load and the fuel economy requirement under full voyage, how does the ECMS energy management strategy based on instantaneous optimization algorithm learn from the global optimal solution determined based on the DP algorithm to achieve fast and reliable online control. Therefore, a new method of fuel cell ship energy management strategy is needed. SUMMARY

[0004] Therefore, the method based on the rule, global optimization and instantaneous optimization strategy cannot effectively solve the multi-power energy management strategy problem of the ship with high fluctuation load. The purpose of the present application is to provide a method for fuel cell ship energy management strategy based on adaptive neural fuzzy inference, which can not only refer to the global optimal result of dynamic programming (DP) algorithm, but also has the ability of online calculation through the trained adaptive neural fuzzy inference system (ANFIS) network, so as to improve the equivalent factor calculation of traditional ECMS to improve the efficiency.

[0005] To achieve the above purpose, the present application adopts the following technical solutions: an improved method for calculating ECMS equivalent factor by using ANFIS to realize online ideal power distribution, comprising the following steps:

[0006] Step S1: obtaining the topological structure of the fuel cell ship hybrid power system and the information of its typical operating conditions;

[0007] Step S2: according to the information, calculating the optimal equivalent factor of each moment of the working condition by discrete solution, and taking the working condition calculation result as a training sample;

[0008] Step S3: building and training ANFIS to calculate the optimal equivalent factor, and bringing the result into the ECMS algorithm to calculate the optimal distribution result of the ship navigation condition online.

[0009] The step S1 comprises the following steps:

[0010] Step S1: obtaining the topological structure of the fuel cell ship hybrid power system and the information of its typical operating conditions;

[0011] Step S11: a fuel cell ship composed of a proton exchange membrane fuel cell, a battery, a super capacitor and a propulsion system;

[0012] The fixed working condition and driving state of the offshore ship specifically include the working condition information of the ship, the navigation parameters of the ship and the load power demand of the ship.

[0013] In view of the power adaptability of the fuel cell ship to high fluctuation load and the fuel economy requirement under the whole voyage, according to the specific power variation amplitude and range, the ship load working condition can be divided into two types: constant speed cruise and range increasing mode. The constant speed cruise is defined as the ship load working condition change stable, and the fuel cell is used as the main power for navigation condition. The range increasing mode is defined as the ship load working condition change fluctuation and instability under the ship starting or porting operation condition, and the fuel cell is used as the range increasing mode working condition.

[0014] The ship hybrid fuel cell power system is composed of a fuel cell energy system, a lithium battery energy system and a super capacitor energy system.

[0015] In order to improve the service life of the battery, a wavelet change power distribution method is adopted under the premise of meeting the load power. The load power is decomposed and reconstructed by three layers of Haar wavelet transform, the reconstructed high frequency power is output as the super capacitor, and the low frequency power is output as the fuel cell and lithium battery. The function of Haar wavelet transform is:

[0016]

[0017] The equivalent hydrogen consumption of the fuel cell composite energy source system is taken as the objective function of the DP algorithm, which is composed of the direct hydrogen consumption of the fuel cell and the equivalent hydrogen consumption of the lithium battery. The real-time equivalent hydrogen consumption expression of each stage is as follows:

[0018] g(t)=C fc (t)+βC bat (t)

[0019] Where g(t) is the equivalent hydrogen consumption at stage t. C fc and C bat are the equivalent hydrogen consumptions of the fuel cell and the lithium battery;

[0020] In order to ensure that the SOC of the lithium battery and the output power of the fuel cell and the lithium battery are kept within reasonable limits, the boundary constraint conditions are as follows:

[0021]

[0022] In the formula: P bat,min and P bat,max are the minimum and maximum values of the lithium battery output power; P fc,min and P fc,max are the minimum and maximum values of the fuel cell output power. The maximum value of the SOC of the lithium battery is 70%, and the minimum value is 50%.

[0023] When solving the multi-stage decision problem, the DP algorithm inversely recursively through the basic recursive formula, and the state value at stage t can be obtained by the state transition equation based on the state at stage t+1. The initial and final SOCs of the lithium battery are controlled to be 60%, and the state transition equation is as follows:

[0024] SOC(t+1)=f(SOC(t),P fc (t),P bat (t))

[0025] In summary, the cumulative objective function of DP algorithm is:

[0026] J(t) = J(t - 1) + g(t)

[0027] ECMS strategy, as a transient optimization strategy, is derived from PMP strategy, and it converts a global optimization problem into a transient optimization problem at each step by solving the Hamilton function extreme value problem. In mathematics, the hydrogen consumption control problem of hybrid power system can be summarized as

[0028]

[0029] where t0is the initial time, t f is the final time. u(t) and x(t) are control variables and state variables, respectively. The state and control variables are given by

[0030]

[0031] Due to power balance, the more the battery power is consumed, the less the fuel consumption is required. Therefore, fuel consumption can be represented as a function of battery power. On the basis of PMP, the Hamilton function H can be represented by

[0032] H(x, u, λ, t) = C fc + C bat + C SC + λ1ΔSOC bat

[0033] where C fc , C bat , C SC are the hydrogen consumptions of fuel cell, lithium battery and super capacitor, respectively, and λ1and λ2are the adjoint variables. Since the terminal voltage and internal resistance of lithium battery are constant, SOC does not change with Hamilton function, so λ1is a constant, i.e. λ1= c.

[0034] Here, the equivalent factor can be defined as Therefore, the objective function of ECMS can be represented as

[0035]

[0036] Taking the power distribution results of DP algorithm as reference, the optimal equivalent factor of ECMS is calculated by discrete optimization. Within the given range of equivalent factor, the equivalent factor s is discretized and substituted into the hydrogen consumption optimal control formula in the search process from large to small, and when the difference of control variable is the smallest, it is the condition that meets.

[0037] ANFIS is a multi-layer model that can combine fuzzy qualitative method with artificial neural network adaptive ability to facilitate adaptation and learning, and is generally composed of several fuzzy rules.

[0038] (1)

[0039] (2)

[0040] (3)

[0041] (4)

[0042] (5)

[0043] Based on the training results of ANFIS, the optimal equivalent factor is calculated online, and the calculation results of the equivalent factor are brought into the ECMS algorithm to calculate the power distribution. Based on the above, the application aims at the power adaptability of fuel cell ships to high fluctuation load and the fuel economy requirement under the whole voyage, establishes an adaptive fuel cell ship hybrid power system, and combines the calculation characteristics of the three strategies, and proposes an improved method of ANFIS optimization ECMS equivalent factor based on dynamic programming, which can online calculate the optimal solution of the real-time power distribution of the ship, and has good global optimization ability.

[0044] Compared with the prior art, the application has the following beneficial effects: in the process of establishing ANFIS to calculate ECMS equivalent factor to realize online power distribution, the mathematical model of the hybrid fuel cell power system is established by taking the actual ship sailing working condition as a reference, and the optimal control distribution route of the ship hydrogen consumption is calculated by the DP algorithm. On this basis, the adaptive neural fuzzy inference system (ANFIS) is built, the optimal value of the equivalent factor is calculated by taking the global optimal calculation result as the standard, and the optimal value is used as the training data, and the training is applied to the analysis and energy management of the ship working condition in real time. BRIEF DESCRIPTION OF DRAWINGS

[0045] Figure 1 is a schematic diagram of the method flow of the application;

[0046] Figure 2 is a schematic diagram of the topology structure of the fuel cell ship hybrid power system of the application;

[0047] Figure 3 is a typical working condition diagram of a fuel cell ship;

[0048] Figure 4is a three-level Haar wavelet decomposition and reconstruction schematic diagram of the present application;

[0049] Figure 5 is a DP algorithm flow chart based on wavelet transform of the present application;

[0050] Figure 6 is an ECMS power distribution result diagram based on ANFIS optimization of the present application;

[0051] Figure 7 is a lithium battery SOC change schematic diagram of the present application;

[0052] Figure 8 is a lithium battery SOC change schematic diagram of the present application under different working conditions;

[0053] Figure 9 is a method flow chart of the present application based on adaptive neuro-fuzzy inference fuel cell ship energy management strategy. DETAILED DESCRIPTION

[0054] In order to make the purpose, technical scheme and advantages of the present application clearer, the embodiments of the present application will be further described below with reference to the drawings.

[0055] Please refer to Figure 1 , Figure 1 is a method flow chart of the present application;

[0056] The present application provides a method based on adaptive neuro-fuzzy inference fuel cell ship energy management strategy, characterized in that it comprises the following steps:

[0057] Step S1: obtaining the topological structure of the fuel cell ship hybrid power system, and the information of its typical working conditions;

[0058] Step S11: a fuel cell ship composed of a proton exchange membrane fuel cell, a battery, a super capacitor and a propulsion system;

[0059] Please refer to Figure 2 , Figure 2 is a topological structure schematic diagram of the fuel cell ship hybrid power system of the present application;

[0060] The above-mentioned fuel cell ship hybrid power system comprises a proton exchange membrane fuel cell, a battery, a super capacitor and a propulsion system.

[0061] Please refer to Figure 3 , Figure 3 is a typical working condition diagram of the fuel cell ship;

[0062] The typical operating conditions of the aforementioned fuel cell ships can be divided into three parts: range-extending mode when departing port, constant-speed cruise, and range-extending mode when berthing. Constant-speed cruise is defined as a navigation condition where the ship's load conditions change stably, and the fuel cell serves as the primary propulsion. Range-extending mode is defined as an operating condition where, during ship startup or berthing, the ship's required load conditions fluctuate significantly and are unstable, and the fuel cell operates in range-extending mode.

[0063] Step S12: For the fuel cell ship hybrid power system described in S11, the objective function is to minimize the equivalent hydrogen consumption. Wavelet transform is used to realize the role of the supercapacitor as a buffer unit of the hybrid power system. The DP algorithm is used to decompose the objective function and solve the multi-stage decision problem. By constraining it, the optimal power allocation result of the corresponding time series is calculated.

[0064] To improve battery life while meeting load power requirements, a wavelet transform-based power allocation method is employed. The load power is decomposed and reconstructed using a three-layer Haar wavelet transform. The reconstructed high-frequency power is used as the output of the supercapacitor, while the low-frequency power is used as the output of the fuel cell and lithium battery. The Haar wavelet transform function is as follows:

[0065]

[0066] Please refer to Figure 4 , Figure 4 This is a schematic diagram of the three-level Haar wavelet decomposition and reconstruction of the present invention.

[0067] The system equivalent hydrogen consumption of the fuel cell hybrid energy source is used as the objective function of the DP algorithm. The system equivalent hydrogen consumption consists of two parts: the direct hydrogen consumption of the fuel cell and the equivalent hydrogen consumption of the lithium battery. The real-time equivalent hydrogen consumption expression for each stage is as follows.

[0068] First, define the real-time equivalent hydrogen consumption as...

[0069] g(t) = C fc (t)+βC bat (t)

[0070] Where g(t) is the equivalent hydrogen consumption at stage t. C fc and C bat Equivalent hydrogen consumption for fuel cells and lithium batteries;

[0071] The formula for calculating equivalent hydrogen consumption is as follows:

[0072]

[0073] in ξ is the molar mass of hydrogen; F is the Faraday constant; ξ is the calorific value of hydrogen; η ch and ηdis For the charging and discharging efficiency of lithium battery, the calculation formula is as follows:

[0074]

[0075] In order to ensure that the SOC of lithium battery and the output power of fuel cell and lithium battery are kept within reasonable limits, the boundary constraint conditions are as follows:

[0076]

[0077] In the formula: P bat,min and P bat,max are the minimum and maximum values of the output power of lithium battery; P fc,min and P fc,max are the minimum and maximum values of the output power of fuel cell. The maximum value of SOC of lithium battery is 70%, and the minimum value is 50%.

[0078] The DP algorithm solves the multi-stage decision problem by inverse recursion through the basic recursive formula, and converts a multi-stage optimal decision process into multiple single-stage optimal decision processes in the transition of the decision process. The division of stages is according to the time sequence, and the discrete results of the SOC of lithium battery and the output power of fuel cell are taken as the state variables and control variables of the stage. Based on the state of t+1 stage, the state value of t stage can be obtained by solving the state transition equation, and the control of the initial and final SOC of lithium battery is 60%, and the state transition equation is as follows:

[0079] SOC(t+1)=f(SOC(t),P bat (t),P f (t))

[0080] Based on the above, the cumulative objective function of DP algorithm is:

[0081] J(t)=J(t-1)+g(t)

[0082] The flow chart of DP algorithm based on wavelet transform is shown in Figure 5 .

[0083] Step S2: According to the above information, the optimal equivalent factor of each moment of the working condition is solved by discretization, and the calculation result of the working condition is taken as the training sample;

[0084] ECMS strategy as instantaneous optimization strategy, it is derived from PMP strategy, through solving the extreme value problem of Hamilton function, a global optimization problem is transformed into instantaneous optimization problem of each step. In mathematics, the hydrogen consumption control problem of hybrid power system can be summarized as

[0085]

[0086] where t0is the initial time, t f is the final time. u(t) and x(t) are the control and state variables, respectively. The state and control variables are given by

[0087]

[0088] Due to the power balance, the more the battery power is depleted, the less fuel consumption is required. Therefore, the fuel consumption can be expressed as a function of the battery power. On the basis of PMP, the Hamiltonian H can be expressed by

[0089] H(x, u, λ, t) = C fc + C bat + C SC + λ1ΔSOC bat

[0090] where C fc , C bat , C SC are the hydrogen consumptions of the fuel cell, lithium battery and super capacitor, respectively, and λ1and λ2are the co-state variables. Since the terminal voltage and internal resistance of the lithium battery are constant, the SOC does not change with the Hamiltonian, so λ1is a constant, i.e. λ1= c.

[0091] Therefore, the hydrogen consumption optimal control problem of the ship hybrid power system can be expressed as

[0092] u * = argminH(x * ,u, λ * ,t)

[0093] Here, the equivalent factor can be defined as Therefore, the objective function of ECMS can be expressed as

[0094]

[0095] The optimal equivalent factor of ECMS is calculated by discrete optimization based on the power distribution results of DP algorithm. Within the given range of equivalent factor, the equivalent factor s is discretized and substituted into the hydrogen consumption optimal control formula in the search process from large to small, and the control variable difference is the smallest when the condition is met.

[0096] Step S3: An adaptive neuro-fuzzy inference system (ANFIS) is built and trained to calculate the optimal equivalent factor, which is brought into the ECMS algorithm to calculate the optimal distribution results of the ship navigation working condition online.

[0097] ANFIS is a multi-layer model that can combine fuzzy qualitative method with artificial neural network adaptive ability to promote adaptation and learning, generally composed of several fuzzy rules. The adaptive neuro-fuzzy inference system (ANFIS) is built, and the general architecture is composed of five layers with neural structure, and the description of different layers is as follows

[0098] (1)

[0099] (2)

[0100] (3)

[0101] (4)

[0102] (5)

[0103] Based on the training results of ANFIS, the optimal equivalent factor is calculated online, and the power distribution calculation is carried out according to the equivalent factor calculation results. The algorithm is verified by MATLAB simulation, and the fuel cell output power is always stable in the high efficiency interval. When facing large amplitude fluctuation load duty, lithium battery and super capacitor are relied on for supplement and absorption, avoiding the rapid change of fuel cell output power, and ensuring the durability of fuel cell. The power distribution results are shown in Figure 6 , and the SOC change of lithium battery is shown in Figure 7 .

[0104] In an embodiment, it needs to be pointed out that the ship is easily affected by the route environment and the channel during navigation, the working condition changes complexly, and the ship working condition information in the prior art only processes the data of ship heading, speed, load and the like, while the embodiment considers that the ship in actual navigation is different from the frequent start and stop of vehicles, and needs certain guarantee of endurance capability, and the voyage time is relatively long, and when updating the global optimal calculation results, certain calculation resources need to be saved, therefore, the ship working condition is further classified according to the specific power variation amplitude and range based on the fluctuation range of ship load working condition, and the stable range threshold is set combining the current environment data and historical data, in the case that the fluctuation data range is greater than or equal to the stable range threshold, the current mode is determined as the range increasing mode, and the ship is in the start or port operation condition, at this time, the fuel cell operates in the range increasing mode; in the case that the fluctuation data range is less than the stable range threshold, the current mode is determined as the constant speed cruising mode, and the ship current load working condition changes stably, at this time, the fuel cell is used as the main power for navigation working condition. The lithium battery SOC change schematic diagram of the above three parts of typical working conditions of fuel cell ship is shown in Figure 8The simulation results show that when the ship is in the departure range-increasing mode, the initial demand power is not large, the fuel cell provides the main output, and as the demand load becomes large and fluctuates greatly, the SOC of the lithium battery first increases and then decreases; when the ship is in the constant-speed cruising state, the fuel cell and the lithium battery jointly maintain the demand load, so the SOC of the lithium battery decreases in a small range; when the ship is in the port range-increasing mode, the demand load changes greatly, and after the lithium battery meets the corresponding power support, timely charging makes the SOC reach the initial position as much as possible. The algorithm can face different high fluctuation navigation loads, stabilize the power output, make the SOC change range of the lithium battery small, reduce the loss caused by the secondary conversion of power, improve the efficiency of the propulsion system and delay the battery life.

[0105] Compared with the prior art, the present application has the following beneficial effects: in the process of establishing ANFIS to calculate the ECMS equivalent factor to realize online power distribution, the mathematical model of the hybrid fuel cell power system is established by taking the actual ship navigation working condition as a reference, and the optimal control distribution route of the ship hydrogen consumption is calculated by the DP algorithm based on wavelet transform. On this basis, an adaptive neuro-fuzzy inference system (ANFIS) is built, the optimal value of the equivalent factor is calculated discretely and successively by taking the global optimal calculation result as a standard, and is used as training data, and is trained and applied in real time in the analysis and energy management of the ship working condition. The calculation advantages of a variety of algorithms and the learning ability of the rule network are combined, so that the present application has real-time online analysis and calculation ability when facing complex ship navigation working conditions.

[0106] The above is only the preferred embodiment of the present application, and is not used to limit the present application, any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the protection scope of the present application.

Claims

1. A method for energy management strategy of fuel cell ships based on adaptive neurofuzzy inference, characterized in that, Includes the following steps: Step S1: Obtain the topology of the fuel cell ship hybrid power system and information on its typical operating conditions, and obtain the globally optimal load distribution of the hybrid power system through an offline algorithm; The fuel cell marine hybrid power system includes: a proton exchange membrane fuel cell, a storage battery, a propulsion system, and a supercapacitor; Typical operating conditions for fuel cell ships include startup, mooring, navigation, and berthing. Based on the specific power variation range, the ship's operating conditions are divided into two categories: constant speed cruise mode and range-extending mode. The constant speed cruise mode is defined as the navigation condition in which the fuel cell serves as the main propulsion when the ship's load conditions are stable. The range-extending mode is defined as the operating condition in which the fuel cell operates in range-extending mode when the ship is starting up or berthing and the required load conditions fluctuate greatly and are unstable. Step S2: Based on the information, the optimal equivalent factor for each moment of the working condition is solved discretely, and the working condition calculation results are used as training samples. Step S3: Build and train an Adaptive Neural Fuzzy Inference System (ANFIS), calculate the optimal equivalence factor, and input the result into the Equivalent Hydrogen Consumption Minimization (ECMS) algorithm to calculate the optimal allocation result for the ship's navigation conditions online. In the process of power allocation calculation and ship's sea operation condition analysis, the specific constraints of the ship's hybrid power system are as follows: in This represents the charge coefficient of a lithium battery. For lithium battery power, is the fuel cell power, and s is the equivalence factor; The computation process of the Equivalent Hydrogen Consumption Minimization (ECMS) algorithm is derived from the PMP strategy. It transforms a global optimization problem into an instantaneous optimization problem at each step by using the Hamiltonian function extremum problem. For fuel cell ships with multiple energy storage units, including proton exchange membrane fuel cells, batteries, supercapacitors, and propulsion systems, the supercapacitors absorb high-frequency load power through wavelet transform, completing the parameter control calculation of multiple equivalent factors. The objective function of ECMS is given as follows: in, , , These are the hydrogen consumption rates for fuel cells, lithium batteries, and supercapacitors, respectively. and It is the equivalent factor for lithium batteries and supercapacitors.

2. The method for energy management strategy of fuel cell ships based on adaptive neurofuzzy inference according to claim 1, characterized in that: The hybrid power system of the proton exchange membrane fuel cell and the battery adopts an adaptive neuro-fuzzy inference fuel cell ship energy management strategy to achieve power allocation with the minimum equivalent hydrogen consumption. Step S1 includes the following steps: Step S11: A fuel cell ship consisting of a proton exchange membrane fuel cell, a storage battery, and a propulsion system; Step S12: For the fuel cell ship hybrid power system described in S11, with the minimum equivalent hydrogen consumption as the objective function, the DP algorithm is used to decompose the objective function and solve the multi-stage decision problem. By performing constraint processing, the optimal power allocation result for the corresponding time series is calculated.

3. The method for energy management strategy of fuel cell ships based on adaptive neurofuzzy inference according to claim 1, characterized in that: The hybrid power system of the proton exchange membrane fuel cell, battery, and supercapacitor employs an adaptive neuro-fuzzy inference fuel cell ship energy management strategy to achieve power allocation with minimal equivalent hydrogen consumption. Step S1 includes the following steps: Step S11: A fuel cell ship consisting of a proton exchange membrane fuel cell, a storage battery, a supercapacitor, and a propulsion system; Step S12: For the fuel cell ship hybrid power system described in S11, the objective function is to minimize the equivalent hydrogen consumption. Wavelet transform is used to realize the role of the supercapacitor as a buffer unit of the hybrid power system. The DP algorithm is used to decompose the objective function and solve the multi-stage decision problem. By constraining it, the optimal power allocation result of the corresponding time series is calculated.

4. The method for energy management strategy of fuel cell ships based on adaptive neurofuzzy inference according to claim 1, characterized in that: ANFIS is a multi-layered model that combines fuzzy qualitative methods with the adaptive capabilities of artificial neural networks to facilitate adaptation and learning. It consists of several fuzzy rules. The Adaptive Neural Fuzzy Inference System (ANFIS) is constructed using a general architecture composed of five layers with neural structures, described below. （1） （2） （3） （4） （5） 。

Images