A Data-Driven Hybrid Automaton Framework to Modeling Complex Dynamical Systems

Sampled data \mathcal{W}=\{w_1,w_2,\cdots,w_L\} of an unknown dynamical system (?) is a collection of sampled L traces obtained by measurement, where for each trace w_i, i = 1,\ldots,L, is a finite sequence of time steps and data (k_{0,i},d_{0,i}),(k_{1,i},d_{1,i}),\cdots,(k_{M_i,i},d_{M_i,i}) in which

40 k_{\ell,i}\in(0,\infty) and k_{\ell+1,i}=k_{\ell,i}+1, \forall \ell = 0,1,\ldots,M_i,\ \forall i = 1,2,\ldots,L. 40 d_{\ell,i}=[x^\top_{i}(k_{\ell,i}),~u^\top_{i}(k_{\ell,i})]^\top \in \mathbb{R}^{n_x+n_u}, \forall \ell = 0,1,\ldots,M_i, \forall i = 1,2,\ldots,L, where x_{i}(k_{\ell,i}),u_{i}(k_{\ell,i}) denote the state and input of the system at \ellth step for ith trace, respectively.

Given neural network (?) with a bounded input set \mathcal{V}, the following set

is called the reachable set of neural network (?).

A neural-network-based hybrid automaton is defined by a tuple \mathcal{H} \triangleq \left \langle \mathcal{Q},\mathcal{X},init,\mathcal{U},\mathcal{E},g,\mathcal{G},inv,\mathcal{F} \right \rangle in which the components are defined by:

40 Topologies: \mathcal{Q} \triangleq \{q_1,q_2,\ldots,q_N\} is a finite set of topologies.

40 State Variables: \mathcal{X} \subset \mathbb{R}^{n_x} is the set of state variables with the state defined by (q,x)\in \mathcal{Q}\times \mathcal{X}.

40 Initial conditions: \emph{init}\subseteq \mathcal{Q}_0\times \mathcal{X}_{0} is the initial state set, in which \mathcal{Q}_0\subseteq \mathcal{Q} and \mathcal{X}_{0}\subseteq \mathcal{X}.

40 Inputs: \mathcal{U} \triangleq \{u_1,u_2,\ldots,u_M\} is the set of inputs for each topologies.

40 Transitions: \mathcal{E}\subset \mathcal{Q} \times \mathcal{Q} is the set of transitions where a discrete transition from ith topology to jth topology is taking place, i.e., q_i \to q_j, e_{ij}=(q_i,q_j)\in \mathcal{E}.

40 Guard functions: g:\mathcal{E} \to \mathcal{G} is the guard function mapping each transition element e_{ij} to its guard g(e_{ij})\in \mathcal{G}.

40 Guards: \mathcal{G}\subseteq2^{\mathcal{X}} is the guard set which satisfies \forall e_{ij}\in \mathcal{E}, g(e_{ij})\in \mathcal{G}. The guard is satisfied by the state when the hybrid automaton model takes a transition from the current topology to another given topology, i.e., (q_k,x_k)\vDash g(e_{ij}) if and only if q_k=q_i and x_k\in g(e_{ij}).

40 Invariants: inv: \mathcal{Q}\to 2^{\mathcal{X}} is a mapping that assigns an invariant inv(q)\subseteq \mathcal{X} for each topology q\in{\mathcal{Q}}. An invariant is satisfied by all the states of a hybrid automata model for a given topology, i.e., (q,x)\vDash inv(q) if and only if x\in inv(q).

40 Set of Dynamical Description: \mathcal{F} is the set of dynamical description which describes the dynamical process for each given topology q\in \mathcal{Q}. In this paper, neural network \Phi_q(x(k),u(k)) \in \mathcal{F} defines the dynamics for each topology q\in \mathcal{Q} in a given time step k\in[k_1,k_2].

Given a compact set \mathcal{X}\subset\mathbb{R}^{n_x}. A finite collection of sets \mathscr{X}\triangleq\{\mathcal{X}^{(1)},\mathcal{X}^{(2)},\ldots,\mathcal{X}^{(N)}\} is called a partition of \mathcal{X} if (1) \mathcal{X}^{(q)} \subseteq \mathcal{X}, \forall q = 1,\ldots,N; (2) \mathrm{int}(\mathcal{X}^{(q)})\cap \mathrm{int}(\mathcal{X}^{(p)})=\emptyset, \forall q \ne p; (3) \mathcal{X}\subseteq \bigcup_{q=1}^{N} \mathcal{X}^{(i)}. Each elements of \mathcal{X}^{(q)} of partition \mathscr{X} is called a cell.

Given sampled trace w_i in set \mathcal{W}_q, an input-output pair is defined as

where d_{\ell,i}=[x^\top_{i}(k_{\ell,i}),~u^\top_{i}(k_{\ell,i})]^\top is given in Definition 1, and d_{\ell,i} \in \mathcal{W}_q \times \mathcal{U}, x_i(k_{\ell+1},i) \in \mathcal{X}_q. The training data set out of \mathcal{W}_q includes all the input-output pairs and is given as below:

where \ell = 0,1,\ldots,M_i, i = 1,\ldots,L.

Transitions between two topologies are automatically generated by the dynamical description of hybrid automaton, such that \forall p,q=1,2,\ldots,N, \ p\neq q, (p,q):x\in\mathcal{X}^{(p)}, \Phi_p(x,u)\in\mathcal{X}^{(q)}.

Given a neural-network-based hybrid automaton model \mathcal{H} with initial set \mathcal{X}_0 and input set \mathcal{U}, the reachable set at a time instant k is \mathcal{X}_{k} \triangleq \{x(k) \mid x(k)~\mathrm{satisfies}~ \mathcal{H}~\mathrm{and}~x(0) \in \mathcal{X}_{0}\} and the reachable set over time interval [0,k_f] is defined by \mathcal{X}_{[0,k_f]} = \bigcup\nolimits_{s=0}^{k_f}\mathcal{X}_{s}.

For a reachable set \mathcal{X}_{k} of \mathcal{H} intersects with l elements of \mathscr{X}, given a subspace \mathcal{V}_{i,k} from \mathcal{X}_{k} in which \mathcal{V}_{i,k}:\mathcal{V}_{i,k}= (\mathcal{X}_{k}\cap\mathcal{X}^{(m)});~ \cup_{i=1}^l\mathcal{V}_{i,k}=\mathcal{X}_{k},~\forall i=1,\cdots,l,~ \exists m=1,\cdots,N, the process of splitting analysis of the output space \mathcal{V}_{i,k+1} is given by

where the process of obtaining \mathcal{V}_{i,k+1},~\forall i = 1,2,\cdots,N is called Split.

For \mathcal{V}_{i,k+1},~i=1,\cdots,l the output reachable set \mathcal{X}_{k+1} for a neural-network-based hybrid automaton model \mathcal{H} at time step k+1 is given by \mathcal{X}_{k+1}\triangleq \bigcup_{i=1}^{l}\mathcal{V}_{i,k} by which the Combine derives the reachable set at k+1 time instance.