A distributed collaborative control method for fuel cell cluster hybrid power system

Abstract

The application discloses a kind of fuel cell cluster hybrid power system distributed collaborative control method, first, fuel cell cluster is regarded as a whole by distributed algorithm to its characteristic parameter is estimated in real time, and real-time output characteristic parameter is given to upper optimization solution layer;Second, in optimization solution layer, the real-time characteristic parameter output from bottom fuel cell cluster is identified and optimized solution, obtains total reference power of fuel cell cluster;Finally, by the total reference power of optimization solution distributed to each fuel cell power generation unit in cluster through the distributed power distribution algorithm of proposal, and through power control link, the power collaborative control of each unit is realized.This method discards the centralized control structure of traditional control method, realizes the distributed real-time estimation of fuel cell cluster characteristic parameter, and the dimension of optimization problem will not be influenced when cluster scale expands, improves the scalability of fuel cell cluster hybrid power system collaborative control.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of fuel cells, and particularly relates to a distributed collaborative control method for a fuel cell cluster hybrid power system. BACKGROUND

[0002] With the rapid development of the global economy, energy demand is increasing, and traditional fossil energy resources are becoming increasingly depleted. In addition, the extensive use of fossil fuels has led to serious environmental pollution and greenhouse gas emissions, exacerbating global climate change. Although renewable energy sources such as solar and wind energy are developing rapidly, their intermittency and instability limit their large-scale application. Hydrogen energy, as a secondary energy source, can be produced from renewable energy sources through water electrolysis and other methods, thereby achieving energy storage and stable supply. Proton exchange membrane fuel cells can convert hydrogen energy into electrical energy, with characteristics of high efficiency, zero emissions, low noise, low temperature operation, fast start-up, etc., which can significantly reduce carbon emissions in transportation and power generation processes. Fuel cells are not only suitable for vehicles, ships, aircraft and rail transportation vehicles, but also suitable for distributed power generation and backup power systems.

[0003] Due to cost and technology limitations, the output power of proton exchange membrane fuel cells is usually between tens and hundreds of kilowatts. In order to be suitable for high-power application scenarios, some scholars have proposed the concept of fuel cell clusters. Compared with single-stack fuel cells, fuel cell clusters have the advantages of high reliability, high redundancy, high flexibility, high efficiency and low decay cost. Fuel cell clusters are suitable for different scales and types of applications, from small mobile devices to large fixed power stations, which can be designed and deployed using fuel cell clusters.

[0004] However, for traditional centralized control strategies, each fuel cell power generation unit in the fuel cell cluster is a direct participant in the collaborative control of the fuel cell cluster hybrid power system. When the scale of the fuel cell cluster is extremely large, i.e. the cluster contains a large number of fuel cell power generation units, it will bring additional communication and computing burden to the system collaborative control, and even may cause a dimension disaster problem, leading to the unsolvable optimization problem. This also limits the large-scale expansion of the fuel cell cluster. SUMMARY

[0005] In order to solve the above problems, the application proposes a distributed collaborative control method for a fuel cell cluster hybrid power system, which discards the centralized control structure of the traditional control method, realizes the distributed real-time estimation of the characteristic parameters of the fuel cell cluster, and the expansion of the cluster scale does not affect the dimension of the collaborative control optimization problem, reduces the communication and computing burden of the fuel cell cluster hybrid power system, and improves the scalability of the collaborative control of the fuel cell cluster hybrid power system.

[0006] To achieve the above object, the technical scheme adopted by the present application is: a fuel cell cluster hybrid power system distributed collaborative control method, comprising the steps of

[0007] Step 1: For the fuel cell cluster, it is regarded as a whole system to estimate the characteristic parameters in real time by a distributed algorithm, wherein each fuel cell power generation unit in the fuel cell cluster is defined as a master node and a plurality of slave nodes, and the characteristic parameters are output in real time to the upper optimization solving layer;

[0008] Step 2: The distributed real-time estimation algorithm of the characteristic parameters of the fuel cell cluster is established, including the master node characteristic parameter distributed real-time estimation algorithm and the slave node characteristic parameter distributed real-time estimation algorithm, and the information of the characteristic parameters of the fuel cell cluster is summarized in the master node through the communication between the adjacent fuel cell power generation units in the fuel cell cluster;

[0009] Step 3: In the master node, the fuel cell cluster is regarded as a whole to output the characteristic parameters to the upper optimization solving layer in real time, the optimization solving layer realizes the identification of the characteristic parameters output by the bottom fuel cell cluster characteristic parameter distributed real-time estimation algorithm in the master node, and the optimization solving is realized by combining the characteristic parameters of the energy storage device to obtain the total reference power of the fuel cell cluster;

[0010] Step 4: The total reference power of the fuel cell cluster obtained by the optimization solving is distributed to each fuel cell power generation unit in the fuel cell cluster, and the active power collaborative control of each unit is realized through the power control link, while the energy storage device passively undertakes the supplement of the power shortage and the regeneration power recovery according to the power balance relationship.

[0011] Further, in the step 1, for the fuel cell cluster, it is regarded as a whole system, and the running state of the fuel cell cluster is reflected by the characteristic parameters estimated in real time by the distributed algorithm, and the characteristic parameters estimated in real time include the characteristic parameters of each fuel cell power generation unit in the fuel cell cluster;

[0012] Therefore, the estimation target process of the characteristic parameters of the fuel cell cluster includes:

[0013] (a) the average efficiency of the fuel cell cluster

[0014]

[0015] Wherein, n represents the number of fuel cell power generation units in the fuel cell cluster, η fci represents the efficiency of the i-th fuel cell power generation unit in the cluster;

[0016] (b) the total hydrogen consumption rate of the fuel cell cluster

[0017]

[0018] wherein, denotes the hydrogen consumption rate of the i-th fuel cell power unit in the cluster;

[0019] (c) the total output power P of the fuel cell cluster fccs :

[0020]

[0021] wherein, P fci denotes the output power of the i-th fuel cell power unit in the cluster;

[0022] (d) the count of normally operating fuel cell units in the fuel cell cluster:

[0023]

[0024] wherein, N flagi denotes the operating state of the i-th fuel cell power unit in the cluster, N flagi = 1 indicates normal operation, N flagi = 0 indicates fault isolation;

[0025] The above (a) to (d) feature parameter estimation targets include: average type parameters and summation type parameters;

[0026] (1) For average type parameters, a feature parameter distributed real-time estimation is constructed based on an average state observer, and the distributed real-time estimation of the average efficiency of the fuel cell cluster is:

[0027]

[0028] wherein, a ηij denotes a communication weight, and denote the efficiency estimation value of the distributed real-time estimation algorithm of the average efficiency of the i-th and j-th fuel cell power unit in the cluster, respectively;

[0029] (2) For summation type parameters, a feature parameter distributed real-time estimation is constructed based on a summation state observer;

[0030] The distributed real-time estimation of the total hydrogen consumption rate of the fuel cell cluster is:

[0031]

[0032] wherein, b cij denotes a communication weight, μ ci and μ cjindirect estimation state of the total hydrogen consumption rate distributed real-time estimation algorithm for the i-th fuel cell power unit in the cluster, c ci denotes the estimation coefficient, P denotes the total hydrogen consumption rate estimation value of the total hydrogen consumption rate distributed real-time estimation algorithm for the i-th fuel cell power unit in the cluster, N i denotes the count estimation value of the count distributed real-time estimation algorithm for the i-th fuel cell power unit in the cluster;

[0033] The total output power distributed real-time estimation of the fuel cell cluster is:

[0034]

[0035] where b pij denotes the communication weight, μ pi and μ pj denotes the indirect estimation state of the total output power distributed real-time estimation algorithm for the i-th and j-th fuel cell power units in the cluster, c pi denotes the estimation coefficient, P fccsi denotes the total output power estimation value of the total output power distributed real-time estimation algorithm for the i-th fuel cell power unit in the cluster;

[0036] The count distributed real-time estimation of the normal operating fuel cell units in the fuel cell cluster is:

[0037]

[0038] where b Nij denotes the communication weight, μ Ni and μ Nj denotes the indirect estimation state of the count distributed real-time estimation algorithm for the i-th and j-th fuel cell power units in the cluster, c Ni denotes the estimation coefficient, N i denotes the count estimation value of the count distributed real-time estimation algorithm for the i-th fuel cell power unit in the cluster.

[0039] Further, each fuel cell power unit in the fuel cell cluster is defined as a master node and a number of slave nodes, for the sum type distributed real-time estimation algorithm, the estimation coefficient is not zero at the master node and the estimation coefficient is zero at the slave node, i.e.:

[0040] For the total hydrogen consumption rate distributed real-time estimation of the fuel cell cluster:

[0041]

[0042] For the total output power distributed real-time estimation of the fuel cell cluster:

[0043]

[0044] For the count distributed real-time estimation of the normal operating fuel cell units in the fuel cell cluster:

[0045]

[0046] Further, in step 2, by the two-by-two communication between the adjacent fuel cell power generation units in the fuel cell cluster, a ring communication network topology is formed, and the distributed real-time estimation algorithm determines the information collection of the characteristic parameters of the fuel cell cluster in the master node; the slave node implements local estimation for the average type distributed real-time estimation algorithm and transmits the average type parameter estimation value to the adjacent node, and the slave node average type distributed real-time estimation algorithm is consistent with the master node, and implements local estimation for the summation type distributed real-time estimation algorithm and transmits the summation type parameter indirect state estimation value to the adjacent node.

[0047] Further, the slave node summation type distributed real-time estimation includes:

[0048] The total hydrogen consumption rate distributed real-time estimation of the fuel cell cluster in the slave node is:

[0049]

[0050] Wherein, represents the hydrogen consumption rate of the i-th fuel cell power generation unit in the cluster, b cij represents the communication weight, μ ci and μ cj respectively represent the indirect estimation of the total hydrogen consumption rate distributed real-time estimation algorithm of the i-th and j-th fuel cell power generation units in the cluster, N i represents the count estimation value of the count distributed real-time estimation algorithm of the i-th fuel cell power generation unit in the cluster;

[0051] The total output power distributed real-time estimation of the fuel cell cluster in the slave node is:

[0052]

[0053] Wherein, P fci represents the output power of the i-th fuel cell power generation unit in the cluster, b pij represents the communication weight, μ pi and μ pj respectively represent the indirect estimation state of the total output power distributed real-time estimation algorithm of the i-th and j-th fuel cell power generation units in the cluster;

[0054] The count distributed real-time estimation of the normal operating fuel cell units in the fuel cell cluster in the slave node is:

[0055]

[0056] wherein N flagi represents the operating state of the i-th fuel cell power unit in the cluster, N flagi = 1 represents normal operation, N flagi = 0 represents fault isolation; b Nij represents the communication weight, μ Ni and μ Nj respectively represent the indirect estimation state of the counting distributed real-time estimation algorithm of the i-th and j-th fuel cell power unit in the cluster.

[0057] Further, in step 3, the information of the characteristic parameters is completed according to the distributed real-time estimation in the master node, and the characteristic parameters including the average efficiency of the fuel cell cluster, the total hydrogen consumption rate, the total output power and the count of the normally operating fuel cell units in the cluster are output in real time to the upper optimization solving layer as a whole of the fuel cell cluster;

[0058] The polynomial fitting identification is performed on the average efficiency of the fuel cell cluster with respect to the total output power and the total hydrogen consumption rate of the fuel cell cluster with respect to the total output power by using a parameter identification algorithm;

[0059] The maximum and minimum limits of the total output power of the fuel cell cluster system are obtained according to the count of the normally operating fuel cell units in the cluster;

[0060] The average efficiency of the fuel cell cluster, the total hydrogen consumption rate, the total output power and the count of the normally operating fuel cell units in the cluster, the maximum and minimum limits of the total output power, the total output power change rate limit and the average efficiency curve and the total hydrogen consumption rate curve are transmitted to the optimization solver, and the objective function of the optimization problem with the fuel consumption cost, the system degradation cost or the fuel consumption cost and the system degradation cost as the target and the corresponding constraint conditions are established in the optimization solver, and then the fuel cell cluster total reference power is obtained by combining and optimizing the solution of the energy storage device characteristic parameters.

[0061] Further, the polynomial fitting identification is performed on the average efficiency of the fuel cell cluster with respect to the total output power and the total hydrogen consumption rate of the fuel cell cluster with respect to the total output power by using a parameter identification algorithm, including:

[0062] The curve of the average efficiency of the fuel cell cluster with respect to the total output power is:

[0063]

[0064] wherein e r , e r-1e1, e0 are polynomial fitting coefficients, P fccs Ptotal is the total output power of the fuel cell cluster, r is the polynomial order; f η is the function relationship of the average efficiency about the total output power;

[0065] The curve of the total hydrogen consumption rate of the fuel cell cluster about the total output power is:

[0066]

[0067] where d2, d1, d0 are polynomial fitting coefficients, f c is the function relationship of the total hydrogen consumption rate about the total output power.

[0068] Further, the maximum and minimum limits of the total output power of the fuel cell cluster system are:

[0069]

[0070] where, Pmax, Pmin are the maximum and minimum output power limits of the single fuel cell power generation unit; N i represents the count estimation value of the count distributed real-time estimation algorithm of the i-th fuel cell power generation unit in the cluster;

[0071] and the total output power change rate limit of the fuel cell cluster system is represented as:

[0072]

[0073] where, represents the maximum change rate limit of the output power of the single fuel cell power generation unit.

[0074] Further, in step 4, the total reference power of the fuel cell cluster obtained by the optimization solution needs to be distributed to each fuel cell power generation unit in the fuel cell cluster through the fuel cell cluster distributed power distribution algorithm, and the fuel cell distributed power distribution is:

[0075]

[0076] where, a pij represents the communication weight, P fc_refi and P fc_refj represent the output power reference values of the i-th and j-th fuel cell power generation units in the cluster; due to the difference between the master node and the slave node, P fccs_refi is the total reference power of the fuel cell cluster of the i-th fuel cell power generation unit;

[0077] Specifically, it can be represented as:

[0078]

[0079] P fccs_ref is the total reference power of the fuel cell cluster obtained by optimization solution.

[0080] Further, on the basis of the total reference power of the fuel cell cluster obtained by optimization solution being distributed to each fuel cell power generation unit in the fuel cell cluster through the proposed distributed power distribution algorithm for the fuel cell cluster, active power collaborative control of each unit is realized through the power control link, and the energy storage device passively undertakes the supplement of power shortage and regenerative power recovery according to the power balance relationship.

[0081] The beneficial effects of adopting the technical solution are as follows:

[0082] The application discloses a distributed collaborative control method for a fuel cell cluster hybrid power system. Firstly, the fuel cell cluster is taken as a whole, and distributed algorithms are used to perform real-time estimation on characteristic parameters of the fuel cell cluster, and the characteristic parameters are output to an upper optimization solution layer in real time. Secondly, the real-time characteristic parameters output from the bottom fuel cell cluster are identified and optimized in the optimization solution layer, and total reference power of the fuel cell cluster is obtained. Finally, the total reference power obtained by optimization solution is distributed to each fuel cell power generation unit in the cluster through the proposed distributed power distribution algorithm, and power collaborative control of each unit is realized through the power control link. The method discards the centralized control structure of the traditional control method, realizes distributed real-time estimation of the characteristic parameters of the fuel cell cluster, and does not affect the dimension of the collaborative control optimization problem when the cluster scale expands, thereby improving the scalability of the collaborative control of the fuel cell cluster hybrid power system. In order to remove the limitation of large-scale expansion of the fuel cell cluster from the perspective of the collaborative control strategy, the communication and calculation burden of the fuel cell cluster hybrid power system is reduced, the flexibility and scalability of the system are improved, and the limitation of the fuel cell cluster scale is removed. In addition, the fuel cell cluster can be model-reduced, and the dimension disaster problem in the collaborative control optimization scheme can be effectively avoided.

[0083] The application discloses a distributed real-time estimation method for fuel cell cluster characteristic parameters and a distributed power distribution scheme for fuel cell clusters, which can realize real-time estimation, real-time transmission of the characteristic parameters of the fuel cell cluster, and real-time distribution of the reference power of each fuel cell unit in the fuel cell cluster.

[0084] The collaborative control optimization strategy disclosed by the application requires fuel cell cluster characteristic parameters from the proposed distributed real-time estimation method for fuel cell clusters. Once the collaborative control optimization scheme is designed, it will become a standard paradigm. The change of the fuel cell cluster scale does not involve the change of the dimension of the optimization scheme, which will avoid the dimension disaster problem of the optimization solver and improve the flexibility and scalability of the system. Attached Figure Description

[0085] Figure 1 This is a schematic diagram of a distributed cooperative control method for a fuel cell cluster hybrid power system according to the present invention.

[0086] Figure 2 This is a topology diagram of the fuel cell cluster hybrid power system in an embodiment of the present invention;

[0087] Figure 3 This is a schematic diagram of a distributed real-time estimation algorithm for characteristic parameters of a fuel cell cluster in an embodiment of the present invention;

[0088] Figure 4 This is a schematic diagram of the distributed power allocation algorithm for fuel cell clusters in an embodiment of the present invention;

[0089] Figure 5 This is a control block diagram of the scalable distributed cooperative control method for a fuel cell cluster hybrid power system in an embodiment of the present invention. Detailed Implementation

[0090] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described below with reference to the accompanying drawings.

[0091] In this embodiment, it can be based on, for example, Figure 2 The proposed fuel cell cluster hybrid power system is described. See [link / reference]. Figure 1 As shown, this invention proposes a distributed cooperative control method for a fuel cell cluster hybrid power system, including the following steps:

[0092] Step 1: For the fuel cell cluster, it is treated as a whole system and the characteristic parameters are estimated in real time through a distributed algorithm. Each fuel cell power generation unit in the fuel cell cluster is defined as a master node and several slave nodes, and the characteristic parameters are output to the upper optimization solution layer in real time.

[0093] Step 2: Establish a distributed real-time estimation algorithm for characteristic parameters of fuel cell clusters, including a distributed real-time estimation algorithm for characteristic parameters of master nodes and a distributed real-time estimation algorithm for characteristic parameters of slave nodes. Through mutual communication between adjacent fuel cell power generation units in the fuel cell cluster, the information of characteristic parameters of the fuel cell cluster is summarized at the master node.

[0094] Step 3: On the master node, the fuel cell cluster is treated as a whole and its characteristic parameters are output to the upper optimization solution layer in real time. The optimization solution layer identifies the characteristic parameters output by the distributed real-time estimation algorithm of the characteristic parameters of the underlying fuel cell cluster locally on the master node, and combines them with the characteristic parameters of the energy storage device to perform optimization and solution to obtain the total reference power of the fuel cell cluster.

[0095] Step 4: The total reference power of the fuel cell cluster optimized solution is allocated to each fuel cell generating unit in the fuel cell cluster, and the active power collaborative control of each unit is realized through the power control link, while the energy storage device passively undertakes the supplement of power shortage and the recovery of regenerated power according to the power balance relationship.

[0096] As an optimization scheme of the above embodiment, in the step 1, for the fuel cell cluster, it is regarded as a whole system, and the operating state of the fuel cell cluster is reflected by the real-time estimated characteristic parameters, and the real-time estimated characteristic parameters include the characteristic parameters of each fuel cell generating unit in the fuel cell cluster.

[0097] Therefore, the establishment of the estimation target process of the characteristic parameters of the fuel cell cluster includes:

[0098] (a) The average efficiency of the fuel cell cluster

[0099]

[0100] Wherein, n represents the number of fuel cell generating units in the fuel cell cluster, η fci represents the efficiency of the i-th fuel cell generating unit in the cluster;

[0101] (b) The total hydrogen consumption rate of the fuel cell cluster

[0102]

[0103] Wherein, represents the hydrogen consumption rate of the i-th fuel cell generating unit in the cluster;

[0104] (c) The total output power P fccs of the fuel cell cluster:

[0105]

[0106] Wherein, P fci represents the output power of the i-th fuel cell generating unit in the cluster;

[0107] (d) The count of normally operating fuel cell units in the fuel cell cluster:

[0108]

[0109] Wherein, N flagi represents the operating state of the i-th fuel cell generating unit in the cluster, N flagi = 1 represents normal operation, N flagi = 0 represents fault isolation;

[0110] The feature parameter estimation targets of (a)-(d) above include: average type parameters and summation type parameters;

[0111] (1) For average type parameters, a feature parameter distributed real-time estimation is constructed based on an average state observer, and the distributed real-time estimation of the average efficiency of the fuel cell cluster is: Figure 3

[0112]

[0113] where a ηij represents a communication weight, and respectively represent the efficiency estimation value of the distributed real-time estimation algorithm of the average efficiency of the i-th and j-th fuel cell power generation units in the cluster;

[0114] (3) For summation type parameters, a feature parameter distributed real-time estimation is constructed based on a summation state observer;

[0115] The distributed real-time estimation of the total hydrogen consumption rate of the fuel cell cluster is:

[0116]

[0117] where b cij represents a communication weight, μ ci and μ cj respectively represent the indirect estimation state of the distributed real-time estimation algorithm of the total hydrogen consumption rate of the i-th and j-th fuel cell power generation units in the cluster, c ci represents an estimation coefficient, represents the total hydrogen consumption rate estimation value of the distributed real-time estimation algorithm of the total hydrogen consumption rate of the i-th fuel cell power generation unit in the cluster, N i represents the count estimation value of the count distributed real-time estimation algorithm of the i-th fuel cell power generation unit in the cluster;

[0118] The distributed real-time estimation of the total output power of the fuel cell cluster is:

[0119]

[0120] where b pij represents a communication weight, μ pi and μ pj respectively represent the indirect estimation state of the distributed real-time estimation algorithm of the total output power of the i-th and j-th fuel cell power generation units in the cluster, c pi represents an estimation coefficient, P fccsi represents the total output power estimation value of the distributed real-time estimation algorithm of the total output power of the i-th fuel cell power generation unit in the cluster;

[0121] ​The count distributed real-time estimation of the normal operation fuel cell unit in the fuel cell cluster is:

[0122]

[0123] Wherein, b Nij represents the communication weight, μ Ni and μ Nj respectively represent the indirect estimation state of the count distributed real-time estimation algorithm of the i-th and j-th fuel cell power generation unit in the cluster, c Ni represents the estimation coefficient, N i represents the count estimation value of the count distributed real-time estimation algorithm of the i-th fuel cell power generation unit in the cluster.

[0124] In addition, each fuel cell power generation unit in the fuel cell cluster is defined as a master node and a plurality of slave nodes, and for the summation type distributed real-time estimation algorithm, the estimation coefficient at the master node is not zero, and the estimation coefficient at the slave node is zero, that is:

[0125] For the total hydrogen consumption rate distributed real-time estimation of the fuel cell cluster:

[0126]

[0127] For the total output power distributed real-time estimation of the fuel cell cluster:

[0128]

[0129] For the count distributed real-time estimation of the normal operation fuel cell unit in the fuel cell cluster:

[0130]

[0131] As an optimization scheme of the above embodiment, in step 2, by communicating with each other between adjacent fuel cell power generation units in the fuel cell cluster, a ring communication network topology is formed, and according to the aforementioned fuel cell cluster distributed real-time estimation method, since the summation type distributed real-time estimation algorithm causes the difference between the master node and the slave node, the aforementioned fuel cell cluster distributed real-time estimation algorithm of the application determines that the information of the characteristic parameters of the fuel cell cluster is summarized at the master node; the slave node implements local estimation according to the average type distributed real-time estimation algorithm and transmits the average type parameter estimation value to the adjacent node, and the slave node average type distributed real-time estimation algorithm is consistent with the master node, and for the summation type distributed real-time estimation algorithm, the local estimation is implemented and the summation type parameter indirect state estimation value is transmitted to the adjacent node.

[0132] The slave node summation type distributed real-time estimation algorithm is inconsistent with the master node, and the slave node summation type distributed real-time estimation includes:

[0133] The total hydrogen consumption rate distributed real-time estimation of the fuel cell cluster at the slave node is:

[0134]

[0135] wherein, represents the hydrogen consumption rate of the i-th fuel cell power unit in the cluster, b cij represents the communication weight, μ ci and μ cj respectively represent the indirect estimation of the total hydrogen consumption rate distributed real-time estimation algorithm of the i-th and j-th fuel cell power unit in the cluster, N i represents the count estimation value of the count distributed real-time estimation algorithm of the i-th fuel cell power unit in the cluster;

[0136] The total output power distributed real-time estimation of the fuel cell cluster at the slave node is:

[0137]

[0138] wherein, P fci represents the output power of the i-th fuel cell power unit in the cluster, b pij represents the communication weight, μ pi and μ pj respectively represent the indirect estimation state of the total output power distributed real-time estimation algorithm of the i-th and j-th fuel cell power unit in the cluster;

[0139] The count distributed real-time estimation of the normally operating fuel cell units in the fuel cell cluster at the slave node is:

[0140]

[0141] wherein, N flagi represents the operating state of the i-th fuel cell power unit in the cluster, N flagi = 1 represents normal operation, N flagi = 0 represents fault isolation; b Nij represents the communication weight, μ Ni and μ Nj respectively represent the indirect estimation state of the count distributed real-time estimation algorithm of the i-th and j-th fuel cell power unit in the cluster.

[0142] As an optimization scheme of the above embodiment, in the step 3, the information aggregation of the characteristic parameters is completed at the master node according to the aforementioned distributed real-time estimation, and the fuel cell cluster is taken as a whole to output the characteristic parameters in real time to the upper optimization solution layer, including: the average efficiency of the fuel cell cluster, the total hydrogen consumption rate, the total output power, and the count of the normally operating fuel cell units in the cluster.

[0143] Firstly, the polynomial fitting algorithm is used to identify the average efficiency of fuel cell cluster with respect to total output power and total hydrogen consumption rate of fuel cell cluster with respect to total output power;

[0144] Specifically represented as:

[0145] The curve of average efficiency of fuel cell cluster with respect to total output power is:

[0146]

[0147] Wherein, e r , e r-1 , …, e1, e0 are polynomial fitting coefficients, P fccs is the total output power of fuel cell cluster, r is the polynomial order; f η is the function relationship of average efficiency with respect to total output power;

[0148] The curve of total hydrogen consumption rate of fuel cell cluster with respect to total output power is:

[0149]

[0150] Wherein, d2, d1, d0 are polynomial fitting coefficients, f c is the function relationship of total hydrogen consumption rate with respect to total output power.

[0151] In addition, according to the count of normal operation fuel cell units in the cluster, the maximum and minimum limits of total output power of fuel cell cluster system are obtained, which are represented as:

[0152]

[0153] Wherein, are the maximum and minimum output power limits of single fuel cell power generation unit respectively; N i represents the count of the i-th fuel cell power generation unit in the cluster, which is the count estimation value of the count estimation algorithm of distributed real-time estimation;

[0154] And the total output power change rate limit of fuel cell cluster system is represented as:

[0155]

[0156] Wherein, represents the maximum change rate limit of output power of single fuel cell power generation unit.

[0157] Then, the average efficiency, total hydrogen consumption rate, total output power of the fuel cell cluster, the count of normal operation fuel cell units in the cluster, the maximum and minimum limit of total output power, the total output power change rate limit, and the average efficiency curve and total hydrogen consumption rate curve are transmitted to the optimization solver, and the objective function of the optimization problem with fuel consumption cost, system degradation cost, or both fuel consumption cost and system degradation cost as the target and the corresponding constraint conditions are established in the optimization solver, and then the fuel cell cluster total reference power is obtained by combining the optimization solution with the energy storage device characteristic parameters.

[0158] The steps 3 are all completed locally in the master node.

[0159] As an optimization scheme of the above embodiment, in step 4, the fuel cell cluster total reference power obtained by the optimization solution needs to be distributed to each fuel cell power generation unit in the fuel cell cluster through the proposed fuel cell cluster distributed power distribution algorithm. Specifically, as shown in the formula (1), the fuel cell distributed power distribution is: Figure 4

[0160]

[0161] wherein a pij represents the communication weight, P fc_refi and P fc_refj represent the output power reference values of the i th and j th fuel cell power generation units in the cluster; due to the difference between the master node and the slave node, P fccs_refi is the fuel cell cluster total reference power of the i th fuel cell power generation unit;

[0162] Specifically, it can be represented as:

[0163]

[0164] wherein P fccs_ref is the fuel cell cluster total reference power obtained by the optimization solution.

[0165] On the basis of distributing the fuel cell cluster total reference power obtained by the optimization solution to each fuel cell power generation unit in the fuel cell cluster through the proposed fuel cell cluster distributed power distribution algorithm, active power collaborative control of each unit is realized through the power control link, and the energy storage device passively undertakes the supplement of power shortage and the recovery of regenerated power according to the power balance relationship.

[0166] As shown in the formula (1), the fuel cell distributed power distribution is: Figure 5 The control block diagram of the expandable distributed collaborative control method of the fuel cell cluster hybrid power system in the embodiment is shown in the formula (1), specifically, Figure 5 (a) is the structure of the master node fuel cell power generation unit controller in the present application, Figure 5 ​(b) is a node fuel cell power plant controller architecture of the present invention.

[0167] The foregoing is considered as illustrative only of the principles of the application and the forms thereof which are demonstrated by the described embodiments. Further, those skilled in the art will readily recognize various modifications and changes which can be made to the application without departing from the scope thereof. It is intended that the application be limited only by the scope of the appended claims, including the full scope of equivalents thereof.

Claims

1. A distributed cooperative control method for a fuel cell cluster hybrid power system, characterized by, The method comprises the steps of: Step 1: Real-time estimation of characteristic parameters of the fuel cell cluster by a distributed algorithm, wherein each fuel cell generating unit in the fuel cell cluster is defined as a master node and a plurality of slave nodes, and the characteristic parameters are output in real time to an upper optimization solving layer; Step 2: Establishing a distributed real-time estimation algorithm for the characteristic parameters of the fuel cell cluster, including a master node characteristic parameter distributed real-time estimation algorithm and a slave node characteristic parameter distributed real-time estimation algorithm, and realizing information collection of the characteristic parameters of the fuel cell cluster at the master node through communication between adjacent fuel cell generating units in the fuel cell cluster; Step 3: At the master node, the fuel cell cluster is regarded as a whole to output the characteristic parameters to the upper optimization solving layer, and the optimization solving layer identifies the characteristic parameters output by the distributed real-time estimation algorithm of the bottom fuel cell cluster, and combines the characteristic parameters of the energy storage device to obtain the total reference power of the fuel cell cluster through optimization solving; Step 4: The total reference power of the fuel cell cluster obtained through optimization solving is distributed to each fuel cell generating unit in the fuel cell cluster, and active power collaborative control of each unit is realized through a power control link, while the energy storage device passively undertakes power shortage compensation and regenerative power recovery according to the power balance relationship.

2. The distributed cooperative control method of a fuel cell cluster hybrid power system according to claim 1, characterized by, In the step 1, the fuel cell cluster is regarded as a whole system, and the operating state of the fuel cell cluster is reflected by the characteristic parameters estimated in real time by the distributed algorithm, and the real-time estimated characteristic parameters include the characteristic parameters of each fuel cell generating unit in the fuel cell cluster; Therefore, the process of establishing the estimation target of the characteristic parameters of the fuel cell cluster comprises: (a) average efficiency of the fuel cell cluster wherein n represents the number of fuel cell power generation units in the fuel cell cluster, η fci ηi represents the efficiency of the i-th fuel cell power generation unit in the cluster; (b) total hydrogen consumption rate of the fuel cell cluster wherein, represents the hydrogen consumption rate of the i-th fuel cell power generation unit in the cluster; (c) the total output power P of the fuel cell cluster fccs : where P fci represents the output power of the i-th fuel cell power unit in the cluster; (d) Counting of the normal operation fuel cell units in the fuel cell cluster: wherein N flagi represents the operating state of the i-th fuel cell power unit in the cluster, N flagi = 1 indicates normal operation, N flagi = 0 indicates fault isolation; The above (a)-(d) characteristic parameter estimation targets include average type parameters and summation type parameters; (1) For the average type parameters, the distributed real-time estimation of the characteristic parameters is constructed based on the average state observer, and the distributed real-time estimation of the average efficiency of the fuel cell cluster is: where a ηij represents the communication weight, and respectively represent the efficiency estimation value of the average efficiency distribution real-time estimation algorithm of the i-th and j-th fuel cell power generation unit in the cluster; (2) For the summation type parameters, the distributed real-time estimation of the characteristic parameters is constructed based on the summation state observer; The distributed real-time estimation of the total hydrogen consumption rate of the fuel cell cluster is: where b cij represents the communication weight, μ ci and μ cj respectively represent the total hydrogen consumption rate distribution of the i-th and j-th fuel cell power generation units in the cluster, c ci represents the estimation coefficient, represents the total hydrogen consumption rate estimation value of the total hydrogen consumption rate distribution of the i-th fuel cell power generation unit in the cluster, N i represents the count estimation value of the count distribution of the i-th fuel cell power generation unit in the cluster; The distributed real-time estimation of the total output power of the fuel cell cluster is: where b pij represents the communication weight, μ pi and μ pj respectively represent the total output power of the i-th and j-th fuel cell power generation units in the cluster, c pi represents the estimation coefficient, P fccsi represents the total output power estimation value of the distributed real-time estimation algorithm for the total output power of the i-th fuel cell power generation unit in the cluster; The distributed real-time estimation of the count of the normal operation fuel cell units in the fuel cell cluster is: where b Nij denotes the communication weight, μ Ni and μ Nj denote the indirect estimation state of the counting distributed real-time estimation algorithm for the i-th and j-th fuel cell power unit in the cluster, c Ni denotes the estimation coefficient, N i denotes the counting estimation value of the counting distributed real-time estimation algorithm for the i-th fuel cell power unit in the cluster.

3. The distributed cooperative control method of a fuel cell cluster hybrid power system according to claim 2, characterized by, Each fuel cell generating unit in the fuel cell cluster is defined as a master node and a plurality of slave nodes, and for the summation type distributed real-time estimation algorithm, the estimation coefficient at the master node is not zero, and the estimation coefficient at the slave node is zero, i.e.: For the distributed real-time estimation of the total hydrogen consumption rate of the fuel cell cluster: For the distributed real-time estimation of the total output power of the fuel cell cluster: For the distributed real-time estimation of the count of the normal operation fuel cell units in the fuel cell cluster:

4. The distributed cooperative control method of a fuel cell cluster hybrid power system according to claim 1, characterized by, In step 2, by the communication between each two adjacent fuel cell generating units in the fuel cell cluster, a ring communication network topology is formed, and a distributed real-time estimation algorithm is used to determine the information collection of the characteristic parameters of the fuel cell cluster in the master node; the slave node uses the average type distributed real-time estimation algorithm to realize local estimation and transmit the average type parameter estimation value to the adjacent node, and the slave node average type distributed real-time estimation algorithm is consistent with the master node, and the slave node uses the summation type distributed real-time estimation algorithm to realize local estimation and transmit the summation type parameter indirect state estimation value to the adjacent node.

5. The distributed collaborative control method of a fuel cell cluster hybrid power system according to claim 4, wherein, The summation type distributed real-time estimation of the slave node includes: The distributed real-time estimation of the total hydrogen consumption rate of the fuel cell cluster in the slave node is: wherein, represents the hydrogen consumption rate of the i-th fuel cell power unit in the cluster, b cij represents the communication weight, μ ci and μ cj respectively represent the total hydrogen consumption rate of the i-th and j-th fuel cell power unit in the cluster, N i represents the count estimate of the i-th fuel cell power unit in the cluster; The distributed real-time estimation of the total output power of the fuel cell cluster in the slave node is: Among them, P fci b represents the output power of the i-th fuel cell power generation unit in the cluster. pij Represents the communication weight, μ pi and μ pj These represent the indirect estimation states of the distributed real-time estimation algorithm for the total output power of the i-th and j-th fuel cell power generation units in the cluster, respectively. The distributed real-time estimation of the count of the normally operating fuel cell units in the fuel cell cluster in the slave node is: where N flagi represents the operating state of the i-th fuel cell power unit in the cluster, N flagi = 1 represents normal operation, N flagi = 0 represents fault isolation; b Nij represents the communication weight, μ Ni and μ Nj respectively represent the indirect estimation state of the i-th and j-th fuel cell power unit in the cluster by the counting distributed real-time estimation algorithm.

6. The distributed cooperative control method of a fuel cell cluster hybrid power system according to claim 1, characterized by, In step 3, the information collection of the characteristic parameters is completed in the master node according to the distributed real-time estimation, and the real-time output of the characteristic parameters of the fuel cell cluster as a whole to the upper optimization solving layer includes: the average efficiency, the total hydrogen consumption rate, the total output power and the count of the normally operating fuel cell units in the fuel cell cluster of the fuel cell cluster; The polynomial fitting identification of the average efficiency of the fuel cell cluster about the total output power and the total hydrogen consumption rate of the fuel cell cluster about the total output power is performed by using the parameter identification algorithm; According to the count of the normally operating fuel cell units in the fuel cell cluster, the maximum and minimum limits of the total output power of the fuel cell cluster system are obtained; The average efficiency, the total hydrogen consumption rate, the total output power and the count of the normally operating fuel cell units in the fuel cell cluster, the maximum and minimum limits of the total output power, the total output power change rate limit and the average efficiency curve and the total hydrogen consumption rate curve of the fuel cell cluster are transmitted to the optimization solver, and the objective function of the optimization problem with the fuel consumption cost, the system degradation cost or the fuel consumption cost and the system degradation cost as the target and the corresponding constraint condition are established in the optimization solver, and then the fuel cell cluster total reference power is obtained by combining and optimizing the solution of the energy storage device characteristic parameters.

7. The distributed collaborative control method of a fuel cell cluster hybrid power system according to claim 6, wherein, The polynomial fitting identification of the average efficiency of the fuel cell cluster about the total output power and the total hydrogen consumption rate of the fuel cell cluster about the total output power is performed by using the parameter identification algorithm, including: The curve of the average efficiency of the fuel cell cluster about the total output power is: where e r , e r-1 ,..., e1, e0are polynomial fitting coefficients, P fccs is the total output power of the fuel cell cluster, and r is the polynomial order; f η is the function relationship of the average efficiency with respect to the total output power; The curve of the total hydrogen consumption rate of the fuel cell cluster about the total output power is: Wherein, d2, d1, d0 are polynomial fitting coefficients, f c is the function relationship of total hydrogen consumption rate with respect to total output power.

8. The distributed cooperative control method of a fuel cell cluster hybrid power system according to claim 6, characterized by, The maximum and minimum limits of the total output power of the fuel cell cluster system are: wherein, Pmax, Pminare the maximum and minimum output power limits of a single fuel cell power unit, respectively; N i N i represents the count estimate of the count distributed real-time estimation algorithm for the i-th fuel cell power unit in the cluster. And the total output power change rate limit of the fuel cell cluster system is: wherein, represents the maximum rate of change of the output power of the single fuel cell power generation unit.

9. The distributed cooperative control method of a fuel cell cluster hybrid power system according to claim 1, characterized by, In step 4, the fuel cell cluster total reference power obtained by the optimization solution needs to be distributed to each fuel cell generating unit in the fuel cell cluster by the fuel cell cluster distributed power distribution algorithm, and the fuel cell distributed power distribution is: wherein a pij represents the communication weight, P fc_refi and P fc_refj respectively represent the output power reference value of the i-th and j-th fuel cell power generation unit in the cluster; due to the difference between the master node and the slave node, P fccs_refi is the total reference power of the fuel cell cluster for the i-th fuel cell power generation unit; Specifically, it can be expressed as: P fccs_ref The total reference power of the fuel cell cluster is obtained by optimizing the solution.

10. The distributed collaborative control method of a fuel cell cluster hybrid power system according to claim 9, wherein The total reference power of the fuel cell cluster obtained by the optimization solution is distributed to each fuel cell power generation unit in the fuel cell cluster by the proposed fuel cell cluster distributed power distribution algorithm, and active power collaborative control of each unit is realized through the power control link, while the energy storage device passively undertakes the supplement of power shortage and regenerative power recovery according to the power balance relationship.

Images