From Foundational Patents to AI-Assisted Control: Four Decades of Permanent Magnet Motor Innovation

Permanent magnet motor technology has evolved through four distinct phases since the earliest retrieved patent filings in 1979, with a clear concentration of activity post-2018 reflecting the acceleration of global electrification investment. The dataset spans jurisdictions including CN, DE, JP, US, EP, IT, KR, and WO, with CN and JP being the most numerically prominent, and covers machine types from the foundational permanent magnet synchronous motor (PMSM) through to 2026-frontier multi-port axial-radial hybrid flux designs.

The early foundational phase (pre-1995) is represented by German experimental patents from 1979–1988 exploring permanent magnet rotor-stator interaction as a mechanical phenomenon, with no power electronics integration. Fuji Electrochemical’s 1982 DE filing established the foundational coupling concept; all filings from this era are uniformly inactive, indicating expired or abandoned early-stage IP.

The development phase (1995–2015) saw the emergence of vector control, dq-axis current command frameworks, and sensorless position estimation. Assignees including Hitachi, Toshiba, Mitsubishi Electric, and Hyundai Motor filed foundational control patents during this window. Hyundai Motor Company’s 2011 DE filing introduced battery-voltage-compensated speed control for hybrid vehicles — a critical enabler for EV integration.

The acceleration phase (2016–2022) produced a pronounced cluster of filings from Chinese universities — Harbin Institute of Technology, Hunan University, Shandong University — and automotive-adjacent suppliers, focusing on MTPA (Maximum Torque Per Ampere), MTPV (Maximum Torque Per Voltage), and flux-weakening optimization. Filing volumes in CN jurisdictions for this period substantially outnumber those in any other single jurisdiction in the dataset.

The Highest-Volume Battleground: MTPA, MTPV, and Loss Minimization Control

Torque optimization control — encompassing Maximum Torque Per Ampere (MTPA), Maximum Torque Per Voltage (MTPV), and loss minimization control (LMC) — is the single highest-volume patent cluster in the permanent magnet motor dataset, dominated by CN-jurisdiction filings from Chinese academic institutions. The core engineering challenge is determining the optimal split between d-axis and q-axis stator current to maximize torque output per unit current, per unit voltage, or per unit loss across the full operating envelope of an IPM or PMSM drive.

Harbin Institute of Technology’s 2021 CN filings on MTPA and MTPV current trajectory search methods use nested current angle and amplitude iteration with nonlinear flux linkage models — an approach that laid the groundwork for the AI-assisted adaptive methods appearing in 2024–2026 filings. Hunan University’s 2019 CN patent combines MTPA curves with voltage and current constraint ellipses to avoid torque oscillation. Zhengzhou University of Light Industry’s 2025 CN filing advances the field by accounting for iron loss in the IPMSM mathematical model under a loss discrimination criteria framework.

An automotive-specific sub-theme has emerged around battery-current-constrained MTPA. Steering Solutions IP Holding Corporation (a Nexteer subsidiary) filed a CN patent in 2021 on battery current limitation for PMSM drivers, while Shanghai Haozheng Electromechanical Equipment’s 2026 CN filing combines formula-based MTPA models with virtual bipolar square-wave signal injection for adaptive online parameter correction — the most recent entrant in this sub-cluster as of the dataset’s coverage.

“Chinese academic institutions — Harbin Institute of Technology, Hunan University, and Zhengzhou University of Light Industry — have established the densest concentration of MTPA/MTPV algorithm IP in this dataset. R&D teams entering the field should map freedom-to-operate carefully against CN-jurisdiction filings in the 2019–2023 window, many of which are active.”

According to WIPO, China has been the world’s largest patent filer since 2019, and the PM motor control cluster in this dataset is consistent with that broader trend — with university-led IP generation representing a significant share of active filings. The concentration of MTPA/MTPV IP in CN-jurisdiction filings from 2019–2023 creates a freedom-to-operate consideration for any commercial entrant developing PMSM drive controllers for EV or industrial applications.

Eliminating the Encoder: Sensorless Control and Position Estimation in PM Motors

Sensorless permanent magnet motor control — eliminating mechanical position sensors to reduce cost, volume, and failure modes — is a technically mature but actively innovating cluster, with Wolong Electric Group emerging as the most concentrated single assignee in this space. The approaches span observer-based methods, signal-injection-based techniques, and high-frequency injection (HFI) for standstill and low-speed operation.

Wolong’s 2022 JP filings cover two complementary approaches: a robust HFI-based startup system for IPM motors that achieves closed-loop startup from any initial rotor state, and a multi-model flux observer with dynamic direct flux control for sensorless operation under parameter variation. Their 2025 DE filing extends this to PM-assisted synchronous reluctance motors (PM-SyR), detecting rotor magnetic polarity via leakage flux path inductance variation in rotor barrier bridges — a technique specifically targeting standstill and low-speed operation where back-EMF-based methods fail.

At the frontier of AI-assisted sensorless control, the University of Alabama’s 2024 US patent deploys three coordinated neural networks within an SVPWM converter framework: a controller network, a parameter estimator, and a combined flux-weakening/MTPA network. This architecture — described in a patent filed with the USPTO — represents a step-change from the model-based observers that dominated the 2015–2022 period.

The sensorless control cluster spans application domains from HVAC fan and blower drives — where elimination of expensive position sensors is critical in cost-sensitive applications — to railway traction (Hitachi’s EP and CN filings) and electric power steering. According to the IEA, electric motors account for approximately 45% of global electricity consumption, making efficiency and reliability improvements in sensorless drive technology a meaningful lever for energy reduction across industrial and transport sectors.

Rotor Architecture and Magnet Configuration: Where Hardware Innovation Concentrates

Physical rotor and magnet architecture innovation in permanent magnet motors is less numerically dominant than control algorithm IP in this dataset but shows more concentrated assignee patterns, with Mando, Samsung, Gree, Toshiba, Mitsubishi Electric, and Nantong University as the primary hardware innovators. The central challenge is maximising flux density while managing thermal stress, demagnetization risk, and manufacturing cost — particularly as rare-earth material supply constraints from Nd-Fe-B and Sm-Co sources remain a strategic concern for the industry.

Segmented multi-material magnets combine high-coercivity rare-earth segments with lower-coercivity zones to balance demagnetization resistance and cost. Mando Corporation’s 2022 DE filing covers rotor arrays with segmented permanent magnets; Gree Electric Appliances’ 2024 EP filing positions variable-coercivity magnets on the d-axis and high-coercivity magnets on the q-axis in a tri-magnet pole configuration; Samsung Electronics’ 2022 EP filing achieves a 3:2 or 3:4 slot-to-pole ratio for optimised flux.

Variable magnetization architectures represent the most strategically significant frontier in hardware innovation. Wisconsin Alumni Research Foundation’s 2024 EP filing pairs a low-coercivity magnet with a high-coercivity magnet in series, enabling current-pulse-driven magnetization level adjustment during operation. Kabushiki Kaisha Toshiba’s 2018 EP filing overlaps variable and fixed magnetic force magnets along the d-axis for pole-count switching. Gree’s 2024 EP tri-magnet configuration advances the concept further with a variable-coercivity center magnet structure. These architectures are covered by patents registered with the EPO and represent a relatively uncrowded IP space compared to MTPA/MTPV control.

At the absolute frontier, Nantong University’s 2026 GB filing on a multi-port axial-radial hybrid flux permanent magnet machine combines axial and radial flux paths within a single machine — a topology relevant to compact high-power-density applications such as electric aircraft and next-generation EV motors. Beihang University’s 2021 CN filing for electric aircraft features axially segmented magnets, spoke-type rotor brackets, and hollow shafts to achieve simultaneous high power density and efficiency targets for electric propulsion.

Magnet Health Monitoring and Remaining Useful Life Prediction: The Newest IP Frontier

Magnet demagnetization — driven by thermal stress, overcurrent events, and aging — is a critical reliability risk for permanent magnet motors in EV traction, railway, and aerospace applications. The magnet health monitoring cluster contains very few filed patents in this dataset as of 2024–2025, making it the most strategically open area for early IP positioning identified in this landscape.

Hitachi’s 2024 EP filing estimates magnet flux and temperature at zero-torque-command states using voltage command and speed data, requiring no voltage or speed sensors. Meidensha Corporation’s 2024 JP filing superimposes a parameter-identification current on the drive current to estimate flux linkage and derive magnet temperature — a technique that avoids interrupting normal motor operation. Both approaches address the fundamental challenge of estimating an internal physical quantity (magnet temperature) from externally measurable electrical signals.

The most forward-looking filing in this cluster is Xiamen King Long United Automotive Industry’s 2025 CN patent, which fuses back-EMF total harmonic distortion (THD) and d-axis inductance change for real-time demagnetization detection and remaining useful life (RUL) prediction. This represents a new reliability engineering layer being added to PM motor systems — driven by EV fleet management needs around warranty and safety obligations. GM Global Technology Operations’ 2024 CN filing computes demagnetization torque capability from flux linkage in real time for propulsion safety management, further establishing the automotive motivation for this cluster.

“Magnet health monitoring and RUL prediction is an emerging IP frontier with very few filed patents in this dataset (2024–2025). This represents an area where early filings can establish durable IP positions, particularly for EV fleet operators and OEMs facing warranty and safety obligations around magnet demagnetization.”

Geographic Concentration, White Spaces, and Strategic Implications for PM Motor IP

The geographic and assignee distribution of permanent magnet motor patents reveals distinct concentration patterns by technical domain — with critical implications for freedom-to-operate analysis, competitive positioning, and R&D investment decisions. China’s CN jurisdiction is the largest single jurisdiction by count, consistent with broader global patent trends tracked by WIPO, while JP and DE jurisdictions host significant industrial and tier-1 supplier activity.

Control algorithm innovation is widely distributed rather than concentrated in a single commercial player. Harbin Institute of Technology, Hunan University, Harbin IT, and Shandong University collectively hold the densest concentration of MTPA/MTPV algorithm IP in the CN jurisdiction, with many filings from the 2019–2023 window still active. Hardware architecture innovation (rotor and magnet design) shows more concentrated patterns around Mando, Samsung, Gree, Toshiba, and Mitsubishi Electric. Sensorless control shows notable concentration at Wolong Electric Group, which has filed aggressively in JP and DE jurisdictions across 2022–2025.

Four strategic white spaces emerge from the dataset. First, variable magnetization motor architectures (Wisconsin Alumni Research Foundation, Toshiba, Gree) remain at an early commercial stage with relatively open IP positions compared to conventional IPM control — presenting opportunity for organisations targeting ultra-wide speed range applications. Second, magnet health monitoring and RUL prediction has very few filed patents in the 2024–2025 window. Third, multi-port axial-radial hybrid flux topologies (Nantong University, 2026) are academic-stage with no commercial assignee concentration. Fourth, AI-integrated real-time control combining neural networks with adaptive signal injection is nascent, with only the University of Alabama (2024) and Shanghai Haozheng (2026) as identified filers.

For organisations conducting patent landscape analysis in PM motor technology, the dataset underscores the importance of jurisdiction-specific freedom-to-operate review — particularly for CN filings in the torque control cluster and JP/DE filings in sensorless control. The PatSnap technology intelligence platform enables real-time monitoring of these clusters as new filings emerge across all eight represented jurisdictions.