System and method for privacy-preserving electric-vehicle charging using artificial intelligence integrated blockchain and homomorphic encryption

Computer-implemented method for privacy-preserving authorization and billing in an electric-vehicle charging infrastructure, the method comprising the steps of:

• receiving, by a computing node having a processor and a local memory linked to a permissioned blockchain network, a charging-initiation request comprising an encrypted-user credential and an encrypted charging-demand record, each encrypted through a homomorphic-encryption process using a user-specific public key;
• extracting, by the processor, encryption-metadata information from the encrypted charging demand record, the encryption-metadata information including a precision-scale indicator, a modulus-index identifier, and a multiplicative-depth indicator, and
• generating, in the local memory, a metadata vector representing an instantaneous encryption-computation state;
• executing, by the processor using instructions stored in the local memory, an artificial intelligence inference process trained on encrypted-operation telemetry to process the metadata vector and to generate a parameter-control profile specifying an adjusted polynomial depth, a ciphertext-packing density, and computation-precision parameters required for a stable noise budget in subsequent encrypted operations;
• performing, by the processor, a homomorphic equality-verification operation between the encrypted-user credential and a corresponding encrypted authorization-token stored in a blockchain ledger, the homomorphic equality-verification operation being executed in accordance with the adjusted polynomial depth, a ciphertext-packing density, and computation precision parameters defined in the parameter-control profile, and
• storing an encrypted authorization outcome in the local memory;
• acquiring, by the computing node, encrypted real-time energy-measurement values from a charging interface;
• segmenting, by the computing node, the encrypted real-time energy-measurement values into discrete time-indexed encrypted blocks;
• executing, by the computing node, an artificial-intelligence pattern-recognition process stored in the local memory to evaluate homomorphic differences between consecutive encrypted blocks and to infer an encrypted energy-consumption progression;
• computing, by the processor, an encrypted billing value by performing homomorphic multiplication between the encrypted energy-consumption progression and an encrypted-tariff vector, the encrypted-tariff vector being periodically updated by a decentralized artificial intelligence optimization routine executed across blockchain validator nodes and accessible to the computing node;
• and
• verifying, by the processor, correctness of the encrypted billing value using a zero-knowledge proof verification validated through a blockchain consensus process;
• recording verified encrypted billing value and verification proof in the blockchain ledger;
• operating a multi-agent optimization protocol in which artificial-intelligence agents resident on different blockchain validator nodes generate localized encrypted gradients representing node-specific latency, throughput, and energy-consumption statistics;
• exchange the encrypted gradients through a homomorphic aggregation routine executed within a consortium ledger;
• update individual configuration parameters based on aggregated optimization results;
• adjust encryption-parameter and consensus-timing settings to maintain balanced computational load across all validator nodes;
• receiving encrypted telemetry streams containing timestamped authorization and billing event markers;
• constructs encrypted temporal feature sets corresponding to inter-block propagation intervals and proof-verification durations;
• executes a recurrent encrypted-sequence model that forecasts near-term delay spikes;
• modifies batch-scheduling parameters or task-allocation priorities for forthcoming encrypted transactions according to a predicted latency pattern;
• and
• updating, in the local memory, a training repository for the artificial-intelligence inference process and the artificial intelligence pattern-recognition process with execution metrics derived from a completed encrypted transaction, wherein the artificial-intelligence inference process generating the parameter-control profile performs dynamic feature extraction by continuously monitoring intermediate ciphertext operation statistics comprising instruction-queue saturation, arithmetic latency variance, and energy-overhead deviation, encoding these into a temporal state vector, and feeding the temporal state vector into a multi-stage inference pipeline in which a first stage performs anomaly detection on abnormal encryption parameters, a second stage performs regression to estimate latency impact for each candidate parameter tuple, and a third stage performs a reinforcement-based selection of a candidate parameter tuple yielding lowest expected execution time subject to a minimum privacy-confidence constraint, the candidate parameter tuple being applied in real time to adjust ciphertext re-linearization frequency and batching depth before initiation of the homomorphic equality-verification operation, and wherein during the homomorphic equality-verification operation, an auxiliary artificial-intelligence supervisory process operates concurrently to monitor incremental ciphertext noise accumulation by:
• computing statistical gradients of noise amplitude over multiple computation cycles,
• forecasting future noise escalation using a temporal predictive model trained on encrypted operation logs, and
• triggering preemptive ciphertext rescaling whenever a predicted escalation slope exceeds a learned threshold;
• wherein the auxiliary artificial-intelligence supervisory process further recalibrates the temporal predictive model at a conclusion of each verification batch by incorporating observed post-operation residual noise into its internal gradient history to maintain adaptive noise control accuracy across successive encrypted authorizations, and wherein the decentralized artificial-intelligence optimization routine updating the encrypted-tariff vector operates as a federated-learning process across the blockchain validator nodes, each blockchain validator node locally computing encrypted gradient updates representing a relationship between local encrypted demand statistics and pricing coefficients, homomorphically aggregating encrypted gradients into a global model without revealing underlying data, verifying gradient integrity through on-chain zero-knowledge proofs of correct update computation, and broadcasting a resulting encrypted coefficient updates to all participating nodes;
• wherein an updated encrypted-tariff vector is then recorded in the blockchain ledger and subsequently used by charging stations for encrypted billing computations.