CN112510701A - Multi-energy ship energy management control device and method - Google Patents

Multi-energy ship energy management control device and method Download PDF

Info

Publication number
CN112510701A
CN112510701A CN202011352050.8A CN202011352050A CN112510701A CN 112510701 A CN112510701 A CN 112510701A CN 202011352050 A CN202011352050 A CN 202011352050A CN 112510701 A CN112510701 A CN 112510701A
Authority
CN
China
Prior art keywords
ship
energy
power
load
value
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202011352050.8A
Other languages
Chinese (zh)
Inventor
戴晓强
沈奎成
刘维亭
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Jiangsu University of Science and Technology
Original Assignee
Jiangsu University of Science and Technology
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Jiangsu University of Science and Technology filed Critical Jiangsu University of Science and Technology
Priority to CN202011352050.8A priority Critical patent/CN112510701A/en
Publication of CN112510701A publication Critical patent/CN112510701A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J3/00—Circuit arrangements for ac mains or ac distribution networks
    • H02J3/003—Load forecast, e.g. methods or systems for forecasting future load demand
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • G06N20/10—Machine learning using kernel methods, e.g. support vector machines [SVM]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/004—Artificial life, i.e. computing arrangements simulating life
    • G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J3/00—Circuit arrangements for ac mains or ac distribution networks
    • H02J3/28—Arrangements for balancing of the load in a network by storage of energy
    • H02J3/32—Arrangements for balancing of the load in a network by storage of energy using batteries with converting means
    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J3/00—Circuit arrangements for ac mains or ac distribution networks
    • H02J3/38—Arrangements for parallely feeding a single network by two or more generators, converters or transformers
    • H02J3/381—Dispersed generators
    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J2203/00—Indexing scheme relating to details of circuit arrangements for AC mains or AC distribution networks
    • H02J2203/20—Simulating, e g planning, reliability check, modelling or computer assisted design [CAD]
    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J2300/00—Systems for supplying or distributing electric power characterised by decentralized, dispersed, or local generation
    • H02J2300/10—The dispersed energy generation being of fossil origin, e.g. diesel generators
    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J2300/00—Systems for supplying or distributing electric power characterised by decentralized, dispersed, or local generation
    • H02J2300/20—The dispersed energy generation being of renewable origin
    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J2300/00—Systems for supplying or distributing electric power characterised by decentralized, dispersed, or local generation
    • H02J2300/20—The dispersed energy generation being of renewable origin
    • H02J2300/22—The renewable source being solar energy
    • H02J2300/24—The renewable source being solar energy of photovoltaic origin
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
    • Y02E10/00—Energy generation through renewable energy sources
    • Y02E10/50—Photovoltaic [PV] energy
    • Y02E10/56—Power conversion systems, e.g. maximum power point trackers

Abstract

The invention discloses a multi-energy ship energy management control device which comprises a ship power parameter acquisition module, a ship load prediction module, an energy management distribution module and a control execution module, wherein the ship power parameter acquisition module acquires power parameters of a multi-energy ship, after the power parameters are preprocessed by the ship load prediction module, a load prediction model is trained, a load prediction data value at the current moment is calculated, the energy management distribution module acquires ship driving control information, a Q-learning algorithm is used for solving a multi-energy power distribution target by combining the load prediction data value at the current moment, and the control execution module acquires a solution, so that various energy distribution of the multi-energy ship is controlled. The invention also provides a corresponding control method. According to the characteristics of the multi-energy power system and the real-time operation condition, the invention reasonably distributes and coordinately controls various energy sources, reduces the voltage and frequency fluctuation of the ship power grid, and enables the electric energy quality of the ship power grid and the economic performance of the ship operation to be optimal.

Description

Multi-energy ship energy management control device and method
Technical Field
The invention relates to the field of ship energy control, in particular to a multi-energy ship energy management control device and method.
Background
Due to the increasing requirements on the power supply quality of ship power stations and the power quality of ship power grids, research on the power supply quality and the power quality of the ship power grids is also a hot direction for the research on ship power systems. Meanwhile, in order to alleviate the energy problem and the environmental pollution problem, the ship with high power supply quality, which is integrated with new energy, becomes an important development direction of the future ship. However, the characteristics of the new energy source, such as contingency and time difference, require the new energy source to be used with an energy storage element. Meanwhile, the energy management mode in modern equipment is usually set manually, and the simplest mode is the mode of manually controlling the starting and stopping of a diesel generator set of a ship and the operation of a converter of an energy storage element. Although the method is simple, the operation cost of the ship is increased, the service life of the energy storage element of the ship is shortened, and the power supply quality of the ship is influenced.
Disclosure of Invention
The purpose of the invention is as follows: the invention provides a simple multi-energy ship energy management control device capable of ensuring high ship power supply quality. The invention also aims to provide a multi-energy ship energy management control method.
The technical scheme is as follows: the invention provides a multi-energy ship energy management control device, comprising:
the ship power parameter acquisition module is used for acquiring power parameters of the multi-energy ship;
the ship load prediction module is used for acquiring the power parameters acquired by the ship power parameter acquisition module to obtain historical load data information of a ship, preprocessing the historical load data information to generate a load prediction training data set, training by using the load prediction training data set to obtain a load prediction model, and calculating a load prediction data value at the current moment;
the energy management distribution module is used for acquiring ship driving control information by using a Q-learning algorithm and distributing power to the ship according to the load prediction data value at the current moment;
and the control execution module is used for adjusting the output power of the multi-energy ship according to the numerical information of the power distribution generated by the energy management distribution module, and controlling the start and stop of the ship diesel generator set and the charge and discharge state of the ship energy storage element.
Further, the power parameters collected by the ship power parameter collecting module include: voltage and current information of a diesel generator and an energy storage element of the ship and residual electric quantity information of the energy storage element.
Further, the preprocessing of the historical load data information by the ship load prediction module comprises the steps of removing bad values from the historical load data information, and predicting the load data of the multi-energy ship by a support vector machine based on particle swarm optimization parameters to obtain a ship load predicted value Y at the moment tt' calculating the actual load demand value of the ship at the time t as YtAnd the predicted value Y of the loadt' calculating the relative error percentage of the difference value, fitting a relative error trend S (t) between the actual value and the load value aiming at the relative error percentage of each time period, and then the load prediction data value at the current time t is as follows:
Y(t)＝(1+S(t))×Yt′
furthermore, the historical load data information is historical actual power demand values of the ship in each time period, and a training data set S { (x) for ship load prediction is obtained after bad values are removed from the historical actual power demand values1,y1),…,(xl,yl)}∈Rn×R，xi∈RnRepresenting inputs of ship electrical load training samples, yiE.g. R represents the corresponding output, and l is the total number of training samples;
establishing an input space RnSet of mappings f onto output R: rn→ R, satisfies f (x) y, then the regression function:
Figure BDA0002801590500000021
calculating the regression error when the support vector machine predicts:
Figure BDA0002801590500000022
the Langrange equation is established as follows:
Figure BDA0002801590500000023
wherein alpha isi,αi *Is a Langcange multiplier;
data points are used to describe ω, specifically:
Figure BDA0002801590500000024
the regression form of the support vector machine for the prediction of the ship power load is as follows:
Figure BDA0002801590500000025
in the above formula, k (x)iX) represents a kernel function, specifically chosen as:
Figure BDA0002801590500000026
wherein σ is the nucleus width;
converting the error problem of the support vector machine load prediction algorithm into the selection problem of the kernel width sigma and the penalty parameter C of the regression error, obtaining the optimized kernel width sigma and the penalty parameter C of the regression error by adopting the particle swarm optimization, and calculating to obtain the ship load prediction data value Y of each time periodt′；
Comparing the ship load prediction data of each time period with the actual ship load demand value, and calculating the prediction relative error percentage of each time period, namely EtThe value of (c):
Figure BDA0002801590500000031
and fitting a relative error trend between the actual value and the load value according to the relative error percentage of each time period:
S(t)＝f(Et)+b
the relative error trend at any time t is S (t), and the predicted value of the ship load is YtFrom this, the predicted value of the ship load at time t +1 can be obtained, that is:
Y(t)＝(1+S(t))×Yt′
further, the energy management distribution module is configured to establish a multi-energy ship power distribution objective function:
Figure BDA0002801590500000032
wherein u (t) and n (t) are control variables and are respectively the output power P of the lithium battery of the multi-energy shipbat(t) output power P of super capacitorsc(t); x (t) is a state variable including the power demand P of the whole shipm(t) SOC value of lithium Batterybat(t) super capacitor SOC value SOCsc(t) and vessel speed v (t); pdie(t) is the output power of the diesel generator, Ppv(t) is the output power of the photovoltaic cell,
Figure BDA0002801590500000034
the actual diesel consumption rate of the diesel generator is shown, alpha is a regulating factor of the charge state of the battery, and beta is a regulating factor of the charge state of the super capacitor;
and solving the multi-energy ship power distribution objective function by adopting a Q-learning algorithm according to the ship load prediction data and the collected ship driving control information to obtain an optimized multi-energy ship power distribution value.
Further, the energy management distribution module adopts a Q-learning algorithm to solve the multi-energy ship power distribution objective function, and includes:
(1) establishing a required power change model:
Figure BDA0002801590500000033
in which the required power is divided into n states, PiIndicating the power P at a certain speediTransfer to PjTransition probability of, NijRepresents from PiTransfer to PjNumber of occurrences, NiRepresents a state PiThe total number of generations;
(2) the energy management problem of the ship power system is described as a quintuple { S, A, { P }, gamma, R }, wherein S is a finite number state set, and parameters in the set comprise required power Pm(t), SOC (t) and navigational speed v (t), wherein A is the output power P of the lithium battery and the super capacitorbat(t)、Psc(t) and P, wherein { P } is a state transition probability distribution adopting the action a in the state s, gamma is a discount factor in the learning process, alpha is a regulation factor of the charge state of the battery, beta is a regulation factor of the charge state of the super capacitor, and R is a return function set expressed by a performance optimization cost function of the ship:
Figure BDA0002801590500000041
(3) for each state s, a value function is defined as the mathematical expectation of the cumulative discount return:
Figure BDA0002801590500000042
where γ is a discount factor representing a discount factor reported over time, E represents an accumulated reward feedback amount, for a fixed policy, and a feedback cost function VπSatisfy the bellman equation:
Figure BDA0002801590500000043
where s' represents the next possible state after the state s has performed the action pi(s), subject to Psπ(s)And (4) distribution. The above formula consists of two parts: real-time rewards R(s) and future cumulative discount reward expectation Es’～Psπ(s)[Vπ(s’)]. Solving for VπThe purpose of the method is to find an optimal action strategy pi(s) in a current state s, and define an optimal value function as:
Figure BDA0002801590500000044
the Bellman equation form is:
Figure BDA0002801590500000045
expressed as the Q function in reinforcement learning, is:
Figure BDA0002801590500000046
wherein Q (s, a) represents the maximum cumulative rebate return for when action a is performed as the first action in the s state; determining an optimal control strategy by solving an optimal value:
Figure BDA0002801590500000047
for the iterative Q-learning algorithm, given the discrete state and motion space, the Q and optimal Q values corresponding to state s and motion a can be defined recursively as:
Figure BDA0002801590500000048
Figure BDA0002801590500000051
finally, the iterative update rule of the Q value in the Q-learning algorithm can be expressed as:
Figure BDA0002801590500000052
where η is the learning rate.
Further, the control execution module is configured to continuously obtain the power parameters of the multi-energy ship collected by the ship power parameter collection module, where the power parameters include output voltage and current information of the multi-energy and remaining power information of the energy storage element; according to the set charging upper limit and the set discharging lower limit of the energy storage element, continuously judging whether the residual electric quantity exceeds the upper limit and the lower limit:
when the residual electric quantity of the energy storage element of the ship is higher than the lower discharge limit, the energy storage element of the ship is used as a main energy source of the ship;
and when the residual electric quantity of the energy storage element of the ship is lower than the lower discharge limit, controlling to start the energy source of the ship, outputting power by the energy source according to the allowance defect value, and switching the energy storage element of the ship into a charging mode.
The invention relates to a multi-energy ship energy management control method, which comprises the following steps:
(1) collecting power parameters of a multi-energy ship as historical load data information;
(2) preprocessing the historical load data information to generate a load prediction training data set, training by using the load prediction training data set to obtain a load prediction model, and calculating a load prediction data value at the current moment;
(3) and acquiring ship driving control information by using a Q-learning algorithm, and performing power distribution on the ship according to the load prediction data value at the current moment.
Further, the step (2) includes:
(21) the actual power demand value of the ship in each historical time period is used as historical load data information, and the historical load data information is obtainedRemoving bad values to obtain a training data set S { (x) for ship load prediction1，y1),…,(xl,yl)}∈Rn×R，xi∈RnRepresenting inputs of ship electrical load training samples, yiE.g. R represents the corresponding output, and l is the total number of training samples;
(22) establishing an input space RnSet of mappings f onto output R: rn→ R, satisfies f (x) y, then the regression function:
Figure BDA0002801590500000053
calculating the regression error when the support vector machine predicts:
Figure BDA0002801590500000054
the Langrange equation is established as follows:
Figure BDA0002801590500000061
wherein alpha isi,αi *Is a Langcange multiplier;
data points are used to describe ω, specifically:
Figure BDA0002801590500000062
obtaining a regression form of a support vector machine for predicting the power load of the ship:
Figure BDA0002801590500000063
in the above formula, k (x)iX) represents a kernel function, specifically chosen as:
Figure BDA0002801590500000064
wherein σ is the nucleus width;
(23) converting the error problem of the support vector machine load prediction algorithm into the selection problem of the kernel width sigma and the penalty parameter C of the regression error, obtaining the optimized kernel width sigma and the penalty parameter C of the regression error by adopting the particle swarm optimization, and calculating to obtain the ship load prediction data value Y of each time periodt′；
(24) Comparing the ship load prediction data of each time period with the actual ship load demand value, and calculating the prediction relative error percentage of each time period, namely EtThe value of (c):
Figure BDA0002801590500000065
and fitting a relative error trend between the actual value and the load value according to the relative error percentage of each time period:
S(t)＝f(Et)+b
the relative error trend at any time t is S (t), and the predicted value of the ship load is YtFrom this, a predicted value of the ship load at time t can be obtained, that is:
Y(t)＝(1+S(t))×Yt′
further, the step (3) includes:
(31) establishing a multi-energy ship power distribution objective function:
Figure BDA0002801590500000066
wherein u (t) and n (t) are control variables and are respectively the output power P of the lithium battery of the multi-energy shipbat(t) output power P of super capacitorsc(t); x (t) is a state variable including the power demand P of the whole shipm(t) SOC value of lithium Batterybat(t) super capacitor SOC value SOCsc(t) and vessel speed v (t); pdie(t) is the output power of the diesel generator,Ppv(t) is the output power of the photovoltaic cell,
Figure BDA0002801590500000071
the actual diesel consumption rate of the diesel generator is shown, alpha is a regulating factor of the charge state of the battery, and beta is a regulating factor of the charge state of the super capacitor;
(32) and solving the multi-energy ship power distribution objective function by adopting a Q-learning algorithm according to the ship load prediction data and the collected ship driving control information to obtain an optimized multi-energy ship power distribution value.
Has the advantages that: compared with the prior art, the invention has the following advantages:
(1) the invention uses a novel medium-voltage direct-current power grid structure in order to overcome the defect of insufficient energy density in an alternating-current power grid structure in a modern ship, uses power electronic equipment to rectify a generator set, and an energy storage element and a photovoltaic panel pass through a current converter to form a direct-current marine power grid, thereby overcoming the defect of insufficient energy density of an alternating-current power grid;
(2) the adopted load prediction algorithm is a load prediction algorithm of a support vector machine, and not only supports linear and nonlinear classification, but also supports linear and nonlinear regression; the ship load has the characteristic of strong randomness, so the ship load is predicted by using a support vector machine load prediction algorithm of particle swarm optimization parameters, and the authenticity of predicted data can be further enhanced by introducing a relative error trend function; on the other hand, the support vector machine load prediction algorithm supports small sample data, can solve the problem of high dimension, and can be well suitable for the characteristics of high randomness and volatility of ship loads;
(3) the adopted energy management power distribution strategy based on emotion learning is an algorithm based on value iteration, a Q-table is constructed by executing states and actions, a value quantity Q value is stored, and then the action is selected according to the Q value to obtain greater benefit, so that an optimal control strategy can be selected from a table of the value quantity Q, balance between consumption of a ship diesel generator set and electric quantity sustainability of an energy storage element is balanced, economic performance of ship operation is further enhanced, service life of equipment is prolonged, matching of power supply of a ship and power demand of the ship is realized, and reliability of a ship power supply course is improved;
(4) the multi-energy ship is integrated with the photovoltaic panel as one of the energy sources of the ship, so that the running economic performance and the environmental protection performance of the ship can be enhanced;
(5) the used energy storage element comprises a lithium battery and a super capacitor, and when the load fluctuation of the ship is large, the voltage fluctuation of a direct current bus of the ship can be better balanced, and the stability of a ship power system is better facilitated.
Drawings
FIG. 1 is a schematic structural diagram of an energy management control device of a multi-energy ship according to the present invention;
FIG. 2 is a flow chart of load prediction model building according to an embodiment of the present invention;
FIG. 3 is a flow chart of the Q-learning algorithm of an embodiment of the present invention;
FIG. 4 is a flowchart illustrating a solving process of a control strategy based on a Q-learning algorithm according to an embodiment of the present invention.
Detailed Description
The technical solution of the present invention is further described below with reference to the accompanying drawings and examples.
Referring to fig. 1, the multi-energy-source ship energy management control device according to the present invention is shown, and the device includes a ship power parameter acquisition module, a ship load prediction module, an energy management distribution module, and a control execution module.
The ship power parameter acquisition module is used for acquiring power parameters of the multi-energy ship, and the power parameters comprise voltage and current information of a diesel generator and an energy storage element of the ship and residual electric quantity information of the energy storage element;
the ship load prediction module is used for acquiring the power parameters acquired by the ship power parameter acquisition module to obtain historical load data information of a ship, and preprocessing the historical load data information: removing bad values from actual power demand values of ships in each historical period to obtain a training data set S for ship load prediction{(x1,y1),…,(xl,yl)}∈Rn×R，xi∈RnRepresenting inputs of ship electrical load training samples, yiE.g. R represents the corresponding output, and l is the total number of training samples;
as shown in FIG. 2, an input space R is establishednSet of mappings f onto output R: rn→ R, satisfies f (x) y, then the regression function:
Figure BDA0002801590500000081
calculating the regression error when the support vector machine predicts:
Figure BDA0002801590500000082
the Langrange equation is established as follows:
Figure BDA0002801590500000083
wherein alpha isi,αi *Is a Langcange multiplier;
data points are used to describe ω, specifically:
Figure BDA0002801590500000084
the regression form of the support vector machine for the prediction of the ship power load is as follows:
Figure BDA0002801590500000085
comparing the ship load prediction data of each time period with the actual ship load demand value, and calculating the prediction relative error percentage of each time period, namely EtThe value of (c):
Figure BDA0002801590500000091
and fitting a relative error trend between the actual value and the load value according to the relative error percentage of each time period:
S(t)＝f(Et)+b
the relative error trend at any time t is S (t), and the predicted value of the ship load is Yt' from this, the predicted ship load data at time t can be obtained, namely:
Y(t)＝(1+S(t))×Yt′
the energy management distribution module is used for establishing a multi-energy ship power distribution objective function:
Figure BDA0002801590500000092
wherein u (t) and n (t) are control variables and are respectively the output power P of the lithium battery of the multi-energy shipbat(t) output power P of super capacitorsc(t); x (t) is a state variable including the power demand P of the whole shipm(t) SOC value of lithium Batterybat(t) super capacitor SOC value SOCsc(t) and vessel speed v (t); pdie(t) is the output power of the diesel generator, Ppv(t) is the output power of the photovoltaic cell,
Figure BDA0002801590500000093
the actual diesel consumption rate of the diesel generator is shown, alpha is a regulating factor of the charge state of the battery, and beta is a regulating factor of the charge state of the super capacitor;
and solving the multi-energy ship power distribution objective function by adopting a Q-learning algorithm according to the ship load prediction data and the collected ship driving control information to obtain an optimized multi-energy ship power distribution value.
The Q-learning algorithm is an algorithm based on value iteration, a Q-table is constructed by executing states and actions, a value quantity Q value is stored, and then the action is selected according to the Q value to obtain a large benefit.
The energy management distribution module adopts a Q-learning algorithm to solve the multi-energy ship power distribution objective function, and comprises the following steps:
(1) establishing a required power change model:
Figure BDA0002801590500000094
in which the required power is divided into n states, PiIndicating the power P at a certain speediTransfer to PjTransition probability of, NijRepresents from PiTransfer to PjNumber of occurrences, NiRepresents a state PiThe total number of generations;
(2) the energy management problem of the ship power system is described as a quintuple { S, A, { P }, gamma, R }, wherein S is a finite number state set, and parameters in the set comprise required power Pm(t), SOC (t) and navigational speed v (t), wherein A is the output power P of the lithium battery and the super capacitorbat(t)、Psc(t) and P, wherein { P } is a state transition probability distribution adopting the action a in the state s, gamma is a discount factor in the learning process, alpha is a regulation factor of the charge state of the battery, beta is a regulation factor of the charge state of the super capacitor, and R is a return function set expressed by a performance optimization cost function of the ship:
Figure BDA0002801590500000101
(3) for each state s, a value function is defined as the mathematical expectation of the cumulative discount return:
Figure BDA0002801590500000102
where γ is a discount factor representing a discount factor reported over time, E represents an accumulated reward feedback amount, for a fixed policy, and a feedback cost function VπSatisfy the bellman equation:
Figure BDA0002801590500000103
where s' represents the next possible state after the state s has performed the action pi(s), subject to Psπ(s)And (4) distribution. The above formula consists of two parts: real-time rewards R(s) and future cumulative discount reward expectation Es’～Psπ(s)[Vπ(s’)]. Solving for VπThe purpose of the method is to find an optimal action strategy pi(s) in a current state s, and define an optimal value function as:
Figure BDA0002801590500000104
the Bellman equation form is:
Figure BDA0002801590500000105
expressed as the Q function in reinforcement learning, is:
Figure BDA0002801590500000106
where Q (s, a) represents the maximum cumulative discount return for executing action a as the first action in the s state. Determining an optimal control strategy by solving an optimal value:
Figure BDA0002801590500000107
for the iterative Q-learning algorithm, given the discrete state and motion space, the Q and optimal Q values corresponding to state s and motion a can be defined recursively as:
Figure BDA0002801590500000111
Figure BDA0002801590500000112
finally, the iterative update rule of the Q value in the Q-learning algorithm can be expressed as:
Figure BDA0002801590500000113
where η is the learning rate. Eta belongs to [0,1], the higher the learning rate, the higher the convergence rate, but the overfitting problem is caused, and in the research, the value is set to be 0.1. FIG. 3 shows a calculation process of the Q-learning algorithm in matlab, and the optimal control strategy based on Q-learning is obtained through multiple iterations. The control strategy based on reinforcement learning is to solve the optimal value function under a given discrete state and action space through a Q-learning algorithm based on value iteration, obtain the optimal control strategy through multiple iterations, and the reinforcement learning control strategy solving process is shown in FIG. 4.
The control execution module is used for adjusting the output power of the multi-energy ship according to the numerical information of power distribution generated by the energy management distribution module, controlling the start and stop of the ship diesel generator set, adjusting the output power, the output power of the photovoltaic panel, the charge-discharge state of the ship energy storage element and switching the states of the first, the second and the third load switches, and the specific control strategy comprises the following steps:
(1) the power parameter collector continuously collects output voltage and current information of a diesel generator, a photovoltaic panel, a lithium battery and a super capacitor of the ship, and continuously collects residual electric quantity information of the energy storage element;
(2) and setting the upper limit and the lower limit of the energy storage element, and continuously judging whether the residual electric quantity exceeds the upper limit and the lower limit. When the residual electric quantity of the energy storage element of the ship is higher than the lower discharge limit, the energy storage element of the ship is used as a main energy source of the ship, and the photovoltaic panel is matched with the energy storage element of the ship to supply the energy requirement of the whole ship;
(3) when the residual electric quantity of the energy storage element of the ship is lower than the lower discharge limit, starting a diesel generator of the ship, switching the energy storage element of the ship into a charging mode, and outputting power by the diesel generator of the ship according to the allowance defect value;
(4) when the driving control information of the ship is that the ship slowly runs at a constant speed, the diesel generator of the ship outputs power according to the allowance defect value, and the output power of the photovoltaic panel is adjusted; when the ship runs at full speed, the second-level load and the third-level load of the ship are closed, a diesel engine set of the ship is increased, and the energy storage element is further discharged.
The invention relates to a multi-energy ship energy management control method, which comprises the following steps:
(1) collecting power parameters of a multi-energy ship as historical load data information;
(21) taking the actual power demand values of the ships in each historical time period as historical load data information, and removing bad values to obtain a training data set S { (x) for predicting the ship load1,y1),…,(xl,yl)}∈Rn×R，xi∈RnRepresenting inputs of ship electrical load training samples, yiE.g. R represents the corresponding output, and l is the total number of training samples;
(22) establishing an input space RnSet of mappings f onto output R: rn→ R, satisfies f (x) y, then the regression function:
Figure BDA0002801590500000121
calculating the regression error when the support vector machine predicts:
Figure BDA0002801590500000122
the Langrange equation is established as follows:
Figure BDA0002801590500000123
wherein alpha isi,αi *Is a Langcange multiplier;
data points are used to describe ω, specifically:
Figure BDA0002801590500000124
obtaining a regression form of a support vector machine for predicting the power load of the ship:
Figure BDA0002801590500000125
in the above formula, k (x)iX) represents a kernel function, specifically chosen as:
Figure BDA0002801590500000126
wherein σ is the nucleus width;
(23) converting the error problem of the support vector machine load prediction algorithm into the selection problem of the kernel width sigma and the penalty parameter C of the regression error, obtaining the optimized kernel width sigma and the penalty parameter C of the regression error by adopting the particle swarm optimization, and calculating to obtain the ship load prediction data value Y of each time periodt′；
(24) Comparing the ship load prediction data of each time period with the actual ship load demand value, and calculating the prediction relative error percentage of each time period, namely EtThe value of (c):
Figure BDA0002801590500000127
and fitting a relative error trend between the actual value and the load value according to the relative error percentage of each time period:
S(t)＝f(Et)+b
the relative error trend at any time t is S (t), and the predicted value of the ship load is YtFrom this, a predicted value of the ship load at time t can be obtained, that is:
Y(t)＝(1+S(t))×Yt′
(31) establishing a multi-energy ship power distribution objective function:
Figure BDA0002801590500000131
wherein u (t) and n (t) are control variables and are respectively the output power P of the lithium battery of the multi-energy shipbat(t) output power P of super capacitorsc(t); x (t) is a state variable including the power demand P of the whole shipm(t) SOC value of lithium Batterybat(t) super capacitor SOC value SOCsc(t) and vessel speed v (t); pdie(t) is the output power of the diesel generator, Ppv(t) is the output power of the photovoltaic cell,
Figure BDA0002801590500000132
the actual diesel consumption rate of the diesel generator is shown, alpha is a regulating factor of the charge state of the battery, and beta is a regulating factor of the charge state of the super capacitor;
(32) solving the multi-energy ship power distribution objective function by adopting a Q-learning algorithm according to the ship load prediction data and the collected ship driving control information to obtain an optimized multi-energy ship multi-energy power distribution value:
(321) establishing a required power change model:
Figure BDA0002801590500000133
in which the required power is divided into n states, PiIs shown inPower P at a certain speediTransfer to PjTransition probability of, NijRepresents from PiTransfer to PjNumber of occurrences, NiRepresents a state PiThe total number of generations;
(322) the energy management problem of the ship power system is described as a quintuple { S, A, { P }, gamma, R }, wherein S is a finite number state set, and parameters in the set comprise required power Pm(t), SOC (t) and navigational speed v (t), wherein A is the output power P of the lithium battery and the super capacitorbat(t)、Psc(t) and P, wherein { P } is a state transition probability distribution adopting the action a in the state s, gamma is a discount factor in the learning process, alpha is a regulation factor of the charge state of the battery, beta is a regulation factor of the charge state of the super capacitor, and R is a return function set expressed by a performance optimization cost function of the ship:
Figure BDA0002801590500000134
(323) for each state s, a value function is defined as the mathematical expectation of the cumulative discount return:
Figure BDA0002801590500000141
where γ is a discount factor representing a discount factor reported over time, E represents an accumulated reward feedback amount, for a fixed policy, and a feedback cost function VπSatisfy the bellman equation:
Figure BDA0002801590500000142
where s' represents the next possible state after the state s has performed the action pi(s), subject to Psπ(s)Distributing; the above formula consists of two parts: real-time rewards R(s) and future cumulative discount reward expectation Es’～Psπ(s)[Vπ(s’)](ii) a Solving for VπThe purpose of (A) is to find aAnd (3) defining an optimal action strategy pi(s) under the previous state s as follows:
Figure BDA0002801590500000143
the Bellman equation form is:
Figure BDA0002801590500000144
expressed as the Q function in reinforcement learning, is:
Figure BDA0002801590500000145
where Q (s, a) represents the maximum cumulative discount return for executing action a as the first action in the s state. Determining an optimal control strategy by solving an optimal value:
Figure BDA0002801590500000146
for the iterative Q-learning algorithm, given the discrete state and motion space, the Q and optimal Q values corresponding to state s and motion a can be defined recursively as:
Figure BDA0002801590500000147
Figure BDA0002801590500000148
finally, the iterative update rule of the Q value in the Q-learning algorithm can be expressed as:
Figure BDA0002801590500000149
where η is the learning rate.

Claims (10)

1. A multi-energy source vessel energy management control apparatus, the apparatus comprising:
the ship power parameter acquisition module is used for acquiring power parameters of the multi-energy ship;
the ship load prediction module is used for acquiring the power parameters acquired by the ship power parameter acquisition module to obtain historical load data information of a ship, preprocessing the historical load data information to generate a load prediction training data set, training by using the load prediction training data set to obtain a load prediction model, and calculating a load prediction data value at the current moment;
the energy management distribution module is used for acquiring ship driving control information by using a Q-learning algorithm and distributing power to the ship according to the load prediction data value at the current moment;
and the control execution module is used for adjusting the output power of the multi-energy ship according to the numerical information of the power distribution generated by the energy management distribution module, and controlling the start and stop of the ship diesel generator set and the charge and discharge state of the ship energy storage element.
2. The multi-energy-source ship energy management control device according to claim 1, wherein the power parameters collected by the ship power parameter collecting module comprise: voltage and current information of a diesel generator and an energy storage element of the ship and residual electric quantity information of the energy storage element.
3. The multi-energy source vessel energy management control device according to claim 1, characterized in that: the ship load prediction module preprocesses the historical load data information, namely removing bad values from the historical load data information, and predicting the load data of the multi-energy ship by adopting a support vector machine based on particle swarm optimization parameters to obtain a ship load predicted value Y at the moment tt' calculating the actual load demand value of the ship at the time t as YtAnd load predictionValue Yt' calculating the relative error percentage of the difference value, fitting a relative error trend S (t) between the actual value and the load value aiming at the relative error percentage of each time period, and then the load prediction data value at the current time t is as follows:
Y(t)＝(1+S(t))×Yt′。
4. the multi-energy source vessel energy management control device according to claim 3, characterized in that: the historical load data information is the actual power demand value of the ship in each historical time period, and a training data set S { (x) for predicting the ship load is obtained after bad values are removed from the historical load data information1,y1),…,(xl,yl)}∈Rn×R，xi∈RnRepresenting inputs of ship electrical load training samples, yiE.g. R represents the corresponding output, and l is the total number of training samples;
establishing an input space RnSet of mappings f onto output R: rn→ R, satisfies f (x) y, then the regression function:
Figure FDA0002801590490000011
calculating the regression error when the support vector machine predicts:
Figure FDA0002801590490000012
the Langrange equation is established as follows:
Figure FDA0002801590490000021
wherein alpha isi,αi *Is a Langcange multiplier;
data points are used to describe ω, specifically:
Figure FDA0002801590490000022
the regression form of the support vector machine for the prediction of the ship power load is as follows:
Figure FDA0002801590490000023
in the above formula, k (x)iX) represents a kernel function, specifically chosen as:
Figure FDA0002801590490000024
wherein σ is the nucleus width;
converting the error problem of the support vector machine load prediction algorithm into the selection problem of the kernel width sigma and the penalty parameter C of the regression error, obtaining the optimized kernel width sigma and the penalty parameter C of the regression error by adopting the particle swarm optimization, and calculating to obtain the ship load prediction data value Y of each time periodt′；
Comparing the ship load prediction data of each time period with the actual ship load demand value, and calculating the prediction relative error percentage of each time period, namely EtThe value of (c):
Figure FDA0002801590490000025
and fitting a relative error trend between the actual value and the load value according to the relative error percentage of each time period:
S(t)＝f(Et)+b
the relative error trend at any time t is S (t), and the predicted value of the ship load is YtFrom this, a predicted value of the ship load at time t can be obtained, that is:
Y(t)＝(1+S(t))×Yt′。
5. the multi-energy-source vessel energy management control device according to claim 1, wherein the energy management distribution module is configured to establish a multi-energy-source vessel power distribution objective function:
Figure FDA0002801590490000026
wherein u (t) and n (t) are control variables and are respectively the output power P of the lithium battery of the multi-energy shipbat(t) output power P of super capacitorsc(t); x (t) is a state variable including the power demand P of the whole shipm(t) SOC value of lithium Batterybat(t) super capacitor SOC value SOCsc(t) and vessel speed v (t); pdie(t) is the output power of the diesel generator, Ppv(t) is the output power of the photovoltaic cell,
Figure FDA0002801590490000035
the actual diesel consumption rate of the diesel generator is shown, alpha is a regulating factor of the charge state of the battery, and beta is a regulating factor of the charge state of the super capacitor;
and solving the multi-energy ship power distribution objective function by adopting a Q-learning algorithm according to the ship load prediction data and the collected ship driving control information to obtain an optimized multi-energy ship power distribution value.
6. The multi-energy-source ship energy management control device according to claim 5, wherein the energy management distribution module adopts a Q-learning algorithm to solve the multi-energy-source ship power distribution objective function, and comprises:
(1) establishing a required power change model:
Figure FDA0002801590490000031
in which the required power is divided into n states, PiIndicating the power P at a certain speediTransfer to PjTransition probability of, NijRepresents from PiTransfer to PjNumber of occurrences, NiRepresents a state PiThe total number of generations;
(2) the energy management problem of the ship power system is described as a quintuple { S, A, { P }, gamma, R }, wherein S is a finite number state set, and parameters in the set comprise required power Pm(t), SOC (t) and navigational speed v (t), wherein A is the output power P of the lithium battery and the super capacitorbat(t)、Psc(t) and P, wherein { P } is a state transition probability distribution adopting the action a in the state s, gamma is a discount factor in the learning process, alpha is a regulation factor of the charge state of the battery, beta is a regulation factor of the charge state of the super capacitor, and R is a return function set expressed by a performance optimization cost function of the ship:
Figure FDA0002801590490000032
(3) for each state s, a value function is defined as the mathematical expectation of the cumulative discount return:
Figure FDA0002801590490000033
where γ is a discount factor representing a discount factor reported over time, E represents an accumulated reward feedback amount, for a fixed policy, and a feedback cost function VπSatisfy the bellman equation:
Figure FDA0002801590490000034
where s' represents the next possible state after the state s has performed the action pi(s), subject to Psπ(s)Distributing; the above formula consists of two parts: real-time rewards R(s) and future cumulative discount reward expectation Es,～Psπ(s)[Vπ(s’)](ii) a Solving for VπIs aimed at finding outAn optimal action strategy pi(s) in a current state s, wherein an optimal value function is defined as:
Figure FDA0002801590490000041
the Bellman equation form is:
Figure FDA0002801590490000042
expressed as the Q function in reinforcement learning, is:
Figure FDA0002801590490000043
where Q (s, a) represents the maximum cumulative discount return for executing action a as the first action in the s state. Determining an optimal control strategy by solving an optimal value:
Figure FDA0002801590490000044
for the iterative Q-learning algorithm, given the discrete state and motion space, the Q and optimal Q values corresponding to state s and motion a can be defined recursively as:
Figure FDA0002801590490000045
Figure FDA0002801590490000046
finally, the iterative update rule of the Q value in the Q-learning algorithm can be expressed as:
Figure FDA0002801590490000047
where η is the learning rate.
7. The multi-energy ship energy management control device according to claim 1, wherein the control execution module is configured to continuously obtain the power parameters of the multi-energy ship collected by the ship power parameter collection module, where the power parameters include output voltage of multi-energy, current information, and remaining power information of an energy storage element; according to the set charging upper limit and the set discharging lower limit of the energy storage element, continuously judging whether the residual electric quantity exceeds the upper limit and the lower limit:
when the residual electric quantity of the energy storage element of the ship is higher than the lower discharge limit, the energy storage element of the ship is used as a main energy source of the ship;
and when the residual electric quantity of the energy storage element of the ship is lower than the lower discharge limit, controlling to start the energy source of the ship, outputting power by the energy source according to the allowance defect value, and switching the energy storage element of the ship into a charging mode.
8. A multi-energy ship energy management control method is characterized by comprising the following steps:
(1) collecting power parameters of a multi-energy ship as historical load data information;
(2) preprocessing the historical load data information to generate a load prediction training data set, training by using the load prediction training data set to obtain a load prediction model, and calculating a load prediction data value at the current moment;
(3) and acquiring ship driving control information by using a Q-learning algorithm, and performing power distribution on the ship according to the load prediction data value at the current moment.
9. The multi-energy source vessel energy management control method according to claim 8, wherein the step (2) comprises:
(21) the actual power demand value of the ship in each historical time period is used as historical load data informationAnd removing the bad value from the training data set S { (x) to obtain the ship load prediction1,y1),…,(xl,yl)}∈Rn×R，xi∈RnRepresenting inputs of ship electrical load training samples, yiE.g. R represents the corresponding output, and l is the total number of training samples;
(22) establishing an input space RnSet of mappings f onto output R: rn→ R, satisfies f (x) y, then the regression function:
Figure FDA0002801590490000051
calculating the regression error when the support vector machine predicts:
Figure FDA0002801590490000052
the Langrange equation is established as follows:
Figure FDA0002801590490000053
wherein alpha isi,αi *Is a Langcange multiplier;
data points are used to describe ω, specifically:
Figure FDA0002801590490000054
obtaining a regression form of a support vector machine for predicting the power load of the ship:
Figure FDA0002801590490000055
in the above formula, k (x)iX) represents a kernel function, specifically chosen as:
Figure FDA0002801590490000056
wherein σ is the nucleus width;
(23) converting the error problem of the support vector machine load prediction algorithm into the selection problem of the kernel width sigma and the penalty parameter C of the regression error, obtaining the optimized kernel width sigma and the penalty parameter C of the regression error by adopting the particle swarm optimization, and calculating to obtain the ship load prediction data value Y of each time periodt′；
(24) Comparing the ship load prediction data of each time period with the actual ship load demand value, and calculating the prediction relative error percentage of each time period, namely EtThe value of (c):
Figure FDA0002801590490000061
and fitting a relative error trend between the actual value and the load value according to the relative error percentage of each time period:
S(t)＝f(Et)+b
the relative error trend at any time t is S (t), and the predicted value of the ship load is YtFrom this, a predicted value of the ship load at time t can be obtained, that is:
Y(t)＝(1+S(t))×Yt′。
10. the multi-energy source vessel energy management control method according to claim 8, wherein the step (3) comprises:
(31) establishing a multi-energy ship power distribution objective function:
Figure FDA0002801590490000062
wherein u (t) and n (t) are control variables and are respectively the output power P of the lithium battery of the multi-energy shipbat(t) output work of super capacitorRate Psc(t); x (t) is a state variable including the power demand P of the whole shipm(t) SOC value of lithium Batterybat(t) super capacitor SOC value SOCsc(t) and vessel speed v (t); pdie(t) is the output power of the diesel generator, Ppv(t) is the output power of the photovoltaic cell,
Figure FDA0002801590490000063
the actual diesel consumption rate of the diesel generator is shown, alpha is a regulating factor of the charge state of the battery, and beta is a regulating factor of the charge state of the super capacitor;
(32) and solving the multi-energy ship power distribution objective function by adopting a Q-learning algorithm according to the ship load prediction data and the collected ship driving control information to obtain an optimized multi-energy ship power distribution value.
CN202011352050.8A 2020-11-27 2020-11-27 Multi-energy ship energy management control device and method Pending CN112510701A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011352050.8A CN112510701A (en) 2020-11-27 2020-11-27 Multi-energy ship energy management control device and method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011352050.8A CN112510701A (en) 2020-11-27 2020-11-27 Multi-energy ship energy management control device and method

Publications (1)

Publication Number Publication Date
CN112510701A true CN112510701A (en) 2021-03-16

Family

ID=74966536

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011352050.8A Pending CN112510701A (en) 2020-11-27 2020-11-27 Multi-energy ship energy management control device and method

Country Status (1)

Country Link
CN (1) CN112510701A (en)

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113193554A (en) * 2021-05-06 2021-07-30 上海交通大学 New energy ship energy saving and emission reduction method and system based on data driving
CN113224749A (en) * 2021-04-27 2021-08-06 重庆理工大学 Oil-electricity hybrid energy storage power supply shelter and global optimal discharge control method
CN113239630A (en) * 2021-06-03 2021-08-10 上海交通大学 Wind resource-influenced mobile energy network power generation and voyage optimization method and system
CN113609102A (en) * 2021-08-11 2021-11-05 佛山仙湖实验室 Construction method of energy management database of hybrid drive mining truck
CN114180023A (en) * 2021-12-08 2022-03-15 中国船舶重工集团公司第七一六研究所 Multi-energy ship control management method and device based on load prediction algorithm
CN114301095A (en) * 2021-12-31 2022-04-08 武汉理工大学 PPO2 energy management method based on multi-energy distributed wireless shore power system
CN115456316A (en) * 2022-11-11 2022-12-09 山东交通职业学院 Ship multi-energy power supply system management method
CN116280140A (en) * 2023-04-13 2023-06-23 广东海洋大学 Ship hybrid power energy management method, equipment and medium based on deep learning
CN117559509A (en) * 2024-01-11 2024-02-13 中国海洋大学 Ship Chai Guangchu optimal power matching method and device

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103332284A (en) * 2013-06-27 2013-10-02 上海海事大学 Energy management and control method for electric propulsion system of hybrid power ship
US20170087996A1 (en) * 2015-09-24 2017-03-30 Ship And Ocean Industries R&D Center Energy Management Strategy for Boats and Ships
CN110275440A (en) * 2019-07-02 2019-09-24 江苏科技大学 Multiple-energy-source ship energy management controller and its control method based on load prediction
CN110718940A (en) * 2019-10-11 2020-01-21 江苏科技大学 Multi-energy ship intelligent power distribution method and device based on load prediction

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103332284A (en) * 2013-06-27 2013-10-02 上海海事大学 Energy management and control method for electric propulsion system of hybrid power ship
US20170087996A1 (en) * 2015-09-24 2017-03-30 Ship And Ocean Industries R&D Center Energy Management Strategy for Boats and Ships
CN110275440A (en) * 2019-07-02 2019-09-24 江苏科技大学 Multiple-energy-source ship energy management controller and its control method based on load prediction
CN110718940A (en) * 2019-10-11 2020-01-21 江苏科技大学 Multi-energy ship intelligent power distribution method and device based on load prediction

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
丁明等: "电力系统短期负荷预测软件包的设计与实现", 《合肥工业大学学报（自然科学版）》 *
林歆悠等: "基于增强学习算法的插电式燃料电池电动汽车能量管理控制策略", 《工程科学学报》 *
石晓艳等: "鲶鱼粒子群算法优化支持向量机的短期负荷预测", 《计算机工程与应用》 *

Cited By (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113224749A (en) * 2021-04-27 2021-08-06 重庆理工大学 Oil-electricity hybrid energy storage power supply shelter and global optimal discharge control method
CN113193554B (en) * 2021-05-06 2023-05-23 上海交通大学 New energy ship energy saving and emission reduction method and system based on data driving
CN113193554A (en) * 2021-05-06 2021-07-30 上海交通大学 New energy ship energy saving and emission reduction method and system based on data driving
CN113239630A (en) * 2021-06-03 2021-08-10 上海交通大学 Wind resource-influenced mobile energy network power generation and voyage optimization method and system
CN113609102A (en) * 2021-08-11 2021-11-05 佛山仙湖实验室 Construction method of energy management database of hybrid drive mining truck
CN113609102B (en) * 2021-08-11 2024-03-19 佛山仙湖实验室 Construction method of energy management database of hybrid drive mining truck
CN114180023B (en) * 2021-12-08 2023-09-01 中国船舶集团有限公司第七一六研究所 Multi-energy ship control management method and device based on load prediction algorithm
CN114180023A (en) * 2021-12-08 2022-03-15 中国船舶重工集团公司第七一六研究所 Multi-energy ship control management method and device based on load prediction algorithm
CN114301095A (en) * 2021-12-31 2022-04-08 武汉理工大学 PPO2 energy management method based on multi-energy distributed wireless shore power system
CN114301095B (en) * 2021-12-31 2023-11-17 武汉理工大学 PPO2 energy management method based on multi-energy distributed wireless shore power system
CN115456316A (en) * 2022-11-11 2022-12-09 山东交通职业学院 Ship multi-energy power supply system management method
CN116280140A (en) * 2023-04-13 2023-06-23 广东海洋大学 Ship hybrid power energy management method, equipment and medium based on deep learning
CN116280140B (en) * 2023-04-13 2023-10-10 广东海洋大学 Ship hybrid power energy management method, equipment and medium based on deep learning
CN117559509A (en) * 2024-01-11 2024-02-13 中国海洋大学 Ship Chai Guangchu optimal power matching method and device

Similar Documents

Publication Publication Date Title
CN112510701A (en) Multi-energy ship energy management control device and method
Mirzapour et al. A new prediction model of battery and wind-solar output in hybrid power system
Xiong et al. Towards a smarter hybrid energy storage system based on battery and ultracapacitor-A critical review on topology and energy management
CN106972516B (en) multi-type energy storage multi-stage control method applicable to micro-grid
CN109103912B (en) Industrial park active power distribution system scheduling optimization method considering power grid peak regulation requirements
CN105207259A (en) Energy-management-based micro-grid system dispatching method in grid connection state
CN112491094B (en) Hybrid-driven micro-grid energy management method, system and device
CN111555356B (en) Random variable analysis method for regional comprehensive energy system based on coupled Markov model
Nambisan et al. Optimal energy management of battery supercapacitor aided solar PV powered agricultural feed mill using Pontryagin's minimum principle
Li et al. Online battery protective energy management for energy-transportation nexus
KR20200119367A (en) Demand power prediction device for energy storage system and method for predicting demand power using the same
CN115423153A (en) Photovoltaic energy storage system energy management method based on probability prediction
Xue et al. Adaptive dynamic programming method for optimal battery management of battery electric vehicle
Aslay et al. Machine learning-based estimation of output current ripple in PFC-IBC used in battery charger of electrical vehicles: a comparison of LR, RF and ANN techniques
CN104821619B (en) Renewable energy source-based storage battery charging device and control method thereof
Zahedi et al. Operational strategy optimization of a hybrid green power system based on fuzzy logic controller with considering for optimal sizing and analysis of different priorities for energy storage
CN109347120B (en) Power grid frequency response oriented decision method and system
Roy et al. Cost optimization of battery and Supercapacitor Hybrid Energy Storage System for dispatching solar PV Power
CN115085227A (en) Micro-grid source storage capacity configuration method and device
Valladolid et al. Lithium-ion SOC optimizer consumption using accelerated particle swarm optimization and temperature criterion
CN113555901A (en) Hybrid energy storage capacity optimization method based on improved S-shaped function particle swarm optimization algorithm
CN112467717A (en) Hybrid energy system real-time load distribution method based on fuzzy control
Maslekar et al. Electric Vehicle Impact Analysis in a Microgrid Using Optimized Bio-Inspired Non-Integer Controller
CN112036735A (en) Energy storage capacity planning method and system for energy storage system of photovoltaic power station
Lian et al. Real‐time energy management strategy for fuel cell plug‐in hybrid electric bus using short‐term power smoothing prediction and distance adaptive state‐of‐charge consumption

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
RJ01 Rejection of invention patent application after publication
RJ01 Rejection of invention patent application after publication

Application publication date: 20210316